Support multimethod in export_llama_lib

.ci/scripts/test_lora_multimethod.sh

@@ -0,0 +1,120 @@
+#!/bin/bash
+# Copyright (c) Meta Platforms, Inc. and affiliates.
+# All rights reserved.
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+set -exu
+# shellcheck source=/dev/null
+source "$(dirname "${BASH_SOURCE[0]}")/utils.sh"
+
+cmake_install_executorch_libraries() {
+    echo "Installing libexecutorch.a, libextension_module.so, libportable_ops_lib.a"
+    rm -rf cmake-out
+    cmake --workflow llm-release
+}
+
+cmake_build_llama_runner() {
+    echo "Building llama runner"
+    pushd extension/llm/tokenizers
+    echo "Updating tokenizers submodule"
+    git submodule update --init
+    popd
+    make llama-cpu
+}
+
+cleanup_files() {
+  echo "Deleting downloaded and generated files"
+  rm -rf "${HF_QWEN_PATH}/"
+  rm -rf "${HF_ADAPTER_PATH}/"
+  rm -rf *.pte
+  rm -f result*.txt
+}
+
+# Download LoRA adapter.
+python -m pip install -q huggingface_hub
+HF_ADAPTER_REPO="lucylq/qwen3_06B_lora_math"
+HF_ADAPTER_PATH=$(
+  bash "$(dirname "${BASH_SOURCE[0]}")/download_hf_hub.sh" \
+    --model_id "${HF_ADAPTER_REPO}" \
+    --files "adapter_config.json" "adapter_model.safetensors"
+)
+
+# Download base model (for tokenizer path).
+HF_QWEN_PATH=$(python -c "from huggingface_hub import snapshot_download; print(snapshot_download('unsloth/Qwen3-0.6B'))")
+echo "Model downloaded to: $HF_QWEN_PATH"
+
+### EXPORT MULTIMETHOD PTE ###
+# Set environment variables for OmegaConf interpolation in yaml.
+export LORA_ADAPTER_CHECKPOINT="${HF_ADAPTER_PATH}/adapter_model.safetensors"
+export LORA_ADAPTER_CONFIG="${HF_ADAPTER_PATH}/adapter_config.json"
+
+$PYTHON_EXECUTABLE -m extension.llm.export.export_llm \
+    --config examples/models/qwen3/config/qwen3_multimethod.yaml
+
+### BUILD LLAMA RUNNER ###
+cmake_install_executorch_libraries
+cmake_build_llama_runner
+
+# Runner constants.
+RUNTIME_ARGS="--tokenizer_path=${HF_QWEN_PATH}/ --temperature=0 --seq_len=100 --warmup=1"
+PROMPT="<|im_start|>user Calculate 15% of 80?<|im_end|><|im_start|>assistant"
+
+# Expected outputs.
+EXPECTED_LORA_PREFIX="
+<|im_start|>user Calculate 15% of 80?<|im_end|><|im_start|>assistant
+To calculate 15% of 80"
+
+EXPECTED_BASE_PREFIX="<|im_start|>user Calculate 15% of 80?<|im_end|><|im_start|>assistant:
+<think>
+Okay, so I need to calculate 15% of 80."
+
+### TEST 1: Run lora_forward method ###
+NOW=$(date +"%H:%M:%S")
+echo "Test 1: Multimethod lora_forward. Starting at ${NOW}"
+# shellcheck source=/dev/null
+cmake-out/examples/models/llama/llama_main \
+    --model_path=multimethod_qwen.pte \
+    --method_name=lora_forward \
+    --prompt="${PROMPT}" \
+    ${RUNTIME_ARGS} > result_lora.txt
+NOW=$(date +"%H:%M:%S")
+echo "Finished at ${NOW}"
+
+RESULT=$(cat result_lora.txt)
+if [[ "${RESULT}" == "${EXPECTED_LORA_PREFIX}"* ]]; then
+  echo "Test 1 (lora_forward): Success"
+else
+  echo "Test 1 (lora_forward): Failure"
+  echo "Expected result prefix: ${EXPECTED_LORA_PREFIX}"
+  echo "Actual result: ${RESULT}"
+  cleanup_files
+  exit 1
+fi
+
+### TEST 2: Run base_forward method ###
+NOW=$(date +"%H:%M:%S")
+echo "Test 2: Multimethod base_forward. Starting at ${NOW}"
+# shellcheck source=/dev/null
+cmake-out/examples/models/llama/llama_main \
+    --model_path=multimethod_qwen.pte \
+    --method_name=base_forward \
+    --prompt="${PROMPT}" \
+    ${RUNTIME_ARGS} > result_base.txt
+NOW=$(date +"%H:%M:%S")
+echo "Finished at ${NOW}"
+
+RESULT=$(cat result_base.txt)
+if [[ "${RESULT}" == "${EXPECTED_BASE_PREFIX}"* ]]; then
+  echo "Test 2 (base_forward): Success"
+else
+  echo "Test 2 (base_forward): Failure"
+  echo "Expected result prefix: ${EXPECTED_BASE_PREFIX}"
+  echo "Actual result: ${RESULT}"
+  cleanup_files
+  exit 1
+fi
+
+echo "Multimethod tests passed!"
+cleanup_files

.github/workflows/pull.yml

@@ -826,6 +826,33 @@ jobs:
         # run llama runner in eager mode
         PYTHON_EXECUTABLE=python bash .ci/scripts/test_lora.sh
 
+  test-lora-multimethod-linux:
+    name: test-lora-multimethod-linux
+    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    permissions:
+      id-token: write
+      contents: read
+    strategy:
+      fail-fast: false
+    with:
+      runner: linux.24xlarge
+      docker-image: ci-image:executorch-ubuntu-22.04-clang12
+      submodules: 'recursive'
+      ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
+      timeout: 90
+      script: |
+        # The generic Linux job chooses to use base env, not the one setup by the image
+        CONDA_ENV=$(conda env list --json | jq -r ".envs | .[-1]")
+        conda activate "${CONDA_ENV}"
+
+        PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh --build-tool "cmake"
+
+        # Install llama requirements
+        bash examples/models/llama/install_requirements.sh
+
+        # run llama runner in eager mode
+        PYTHON_EXECUTABLE=python bash .ci/scripts/test_lora_multimethod.sh
+
   test-mediatek-models-linux:
     name: test-mediatek-models-linux
     uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main

examples/models/llama/BUCK

@@ -148,6 +148,7 @@ fbcode_target(_kind = runtime.python_library,
 fbcode_target(_kind = runtime.python_library,
     name = "export_library",
     srcs = [
+        "convert_weights.py",
         "export_llama.py",
         "export_llama_lib.py",
         "model.py",

examples/models/llama/export_llama_lib.py

@@ -16,19 +16,19 @@
 import re
 import shlex
 from functools import partial
-
 from importlib import resources as _resources
 from json import JSONDecodeError
 from pathlib import Path
-from typing import Callable, List, Optional, Union
+from typing import Callable, Dict, List, Optional, Union
 
 import torch
-
 from executorch.devtools.backend_debug import print_delegation_info
 from executorch.devtools.etrecord import generate_etrecord as generate_etrecord_func
 from executorch.examples.models.llama.hf_download import (
     download_and_convert_hf_checkpoint,
 )
+from executorch.exir import to_edge_transform_and_lower
+from executorch.exir.backend.partitioner import Partitioner
 from executorch.exir.passes.init_mutable_pass import InitializedMutableBufferPass
 from executorch.extension.llm.export.builder import DType, LLMEdgeManager
 from executorch.extension.llm.export.config.llm_config import LlmConfig
@@ -52,6 +52,7 @@
 )
 from executorch.util.activation_memory_profiler import generate_memory_trace
 from omegaconf import DictConfig
+from torch.export import ExportedProgram
 
 from ..model_factory import EagerModelFactory
 from .source_transformation.apply_spin_quant_r1_r2 import (
@@ -852,6 +853,28 @@ def _validate_args(llm_config):
                 "Shared embedding is only supported with torchao quantization."
             )
 
+    if llm_config.multimethod_lora.enabled:
+        if llm_config.base.lora_config is not None:
+            raise ValueError(
+                "Cannot use both base.lora_config and multimethod_lora.methods. "
+                "Use multimethod_lora.methods for all LoRA variants."
+            )
+        if llm_config.quantization.pt2e_quantize is not None:
+            raise ValueError(
+                "PT2E quantization is not supported with multimethod_lora export."
+            )
+        if (
+            llm_config.backend.coreml.enabled
+            or llm_config.backend.vulkan.enabled
+            or llm_config.backend.qnn.enabled
+            or llm_config.backend.mps.enabled
+            or llm_config.backend.openvino.enabled
+        ):
+            raise ValueError(
+                "multimethod_lora export only supports XNNPACK backend or portable ops"
+                "Please disable other backends (coreml, vulkan, qnn, mps, openvino)."
+            )
+
 
 def _to_edge_and_lower_llama_xnnpack(
     builder_exported,
@@ -946,7 +969,6 @@ def _to_edge_and_lower_llama_tosa(
     tosa_spec,
     verbose: bool = False,
 ) -> LLMEdgeManager:
-
     logging.info("Lowering model using TOSA partitioner")
 
     partitioners = []
@@ -1141,9 +1163,126 @@ def _to_edge_and_lower_llama(  # noqa: C901
     return builder
 
 
+def _get_xnnpack_partitioners(llm_config: LlmConfig) -> Optional[List[Partitioner]]:
+    """Get XNNPACK partitioners for multimethod_lora export."""
+    partitioners = []
+
+    if llm_config.backend.xnnpack.enabled:
+        partitioners.append(
+            get_xnnpack_partitioner(dynamic_quant_only_partitioner=True)
+        )
+        if llm_config.backend.xnnpack.extended_ops:
+            partitioners.append(
+                get_xnnpack_partitioner(dynamic_quant_only_partitioner=False)
+            )
+
+    return partitioners if partitioners else None
+
+
+def _get_output_filename(
+    llm_config: LlmConfig, modelname: str, output_dir: str, dtype: DType
+) -> str:
+    """Determine output filename for the .pte file."""
+    if dtype == DType.fp16:
+        modelname = f"{modelname}_h"
+
+    if llm_config.export.output_name:
+        output_name = llm_config.export.output_name
+        if output_name.endswith(".pte"):
+            return output_name
+        else:
+            return f"{output_dir}/{output_name}.pte"
+    else:
+        return f"{output_dir}/{modelname}.pte"
+
+
+def _export_llama_multimethod(llm_config: LlmConfig) -> LLMEdgeManager:
+    """
+    Export multiple methods (base + LoRA variants) to a single .pte file.
+
+    For each method in llm_config.multimethod_lora.methods:
+    - If LoraConfig is None: use base model
+    - If LoraConfig is provided: create model with LoRA weights
+
+    Limitations:
+    - Only XNNPACK backend is supported for multimethod_lora export.
+    - PT2E quantization is not supported.
+    - Each method is exported separately; export time scales linearly
+      with the number of methods.
+    - The final .pte file deduplicates shared weights automatically.
+    """
+    num_methods = len(llm_config.multimethod_lora.methods)
+    logging.info(
+        f"multimethod_lora export: exporting {num_methods} method(s). "
+        "Each method requires separate model instantiation and export."
+    )
+
+    additional_passes = []
+    if llm_config.base.model_class.value in TORCHTUNE_DEFINED_MODELS:
+        additional_passes = [InitializedMutableBufferPass(["kv_cache_pos"])]
+
+    # Build dict of exported programs
+    method_to_program: Dict[str, ExportedProgram] = {}
+    first_builder = None
+
+    for method_name, lora_config in llm_config.multimethod_lora.methods.items():
+        logging.info(f"Exporting method: {method_name}")
+
+        # Create a copy of config with this method's LoRA setting
+        method_config = copy.deepcopy(llm_config)
+        method_config.base.lora_config = lora_config
+        # Disable multimethod_lora to avoid infinite recursion
+        method_config.multimethod_lora.methods = {}
+
+        # Load and prepare model for this method
+        builder = _prepare_for_llama_export(method_config)
+        builder = builder.export()
+        builder.run_canonical_optimizations()
+
+        # Get the exported program
+        exported_program = builder._export(builder.pre_autograd_graph_module)
+        method_to_program[method_name] = exported_program
+
+        if first_builder is None:
+            first_builder = builder
+
+    assert first_builder is not None, "No methods to export"
+
+    # Get partitioners based on backend config
+    partitioners = _get_xnnpack_partitioners(llm_config)
+
+    # Lower all methods together using multimethod_lora API
+    edge_config = first_builder._get_edge_config()
+    edge_manager = to_edge_transform_and_lower(
+        method_to_program,
+        partitioner=partitioners,
+        compile_config=edge_config,
+        constant_methods=first_builder.metadata,
+        generate_etrecord=llm_config.debug.generate_etrecord,
+    )
+
+    # Convert to executorch and save
+    first_builder.edge_manager = edge_manager
+    first_builder = first_builder.to_executorch(passes=additional_passes)
+
+    output_file = _get_output_filename(
+        llm_config,
+        first_builder.modelname,
+        first_builder.output_dir,
+        first_builder.dtype,
+    )
+    first_builder.save_to_pte(output_file)
+
+    return first_builder
+
+
 def _export_llama(llm_config: LlmConfig) -> LLMEdgeManager:  # noqa: C901
     _validate_args(llm_config)
 
+    # Check for multimethod_lora export
+    if llm_config.multimethod_lora.enabled:
+        return _export_llama_multimethod(llm_config)
+
     pt2e_quant_params, quantizers, quant_dtype = get_quantizer_and_quant_params(
         llm_config
     )
@@ -1247,23 +1386,12 @@ def _export_llama(llm_config: LlmConfig) -> LLMEdgeManager:  # noqa: C901
     if llm_config.debug.profile_memory:
         generate_memory_trace(builder.export_program, "memory_profile.json")
 
-    if builder.dtype == DType.fp16:
-        modelname = f"{modelname}_h"
-
-    if llm_config.export.output_name:
-        modelname = llm_config.export.output_name
-        if modelname.endswith(".pte"):
-            output_file = modelname
-            modelname = modelname[:-4]
-            print(f"modelname: {modelname}")
-            print(f"output_file: {output_file}")
-        else:
-            output_file = f"{builder.output_dir}/{modelname}.pte"
-            print(f"modelname: {modelname}")
-            print(f"output_file: {output_file}")
-    else:
-        output_file = f"{builder.output_dir}/{modelname}.pte"
-
+    output_file = _get_output_filename(
+        llm_config,
+        modelname,
+        builder.output_dir,
+        builder.dtype,
+    )
     builder.save_to_pte(output_file)
     return builder
 

examples/models/llama/runner/targets.bzl

@@ -47,6 +47,7 @@ def define_common_targets():
                 "//executorch/examples/models/llama/tokenizer:tiktoken",
                 "//pytorch/tokenizers:llama2c_tokenizer",
                 "//pytorch/tokenizers:hf_tokenizer",
+                "//pytorch/tokenizers:regex_lookahead",
             ] + (_get_operator_lib(aten)) + ([
                 # Vulkan API currently cannot build on some platforms (e.g. Apple, FBCODE)
                 # Therefore enable it explicitly for now to avoid failing tests