调试本地部署ai过程中的遇到的问题请求解答

[ddppt-yy]

1、tests目录下的测试脚本在实验过程中发现其使用chat函数似乎参数是缺失的？
`INFO:httpx:HTTP Request: POST http://localhost:11434/api/chat "HTTP/1.1 200 OK"
I am Qwen, an AI designed to assist with answering questions, providing information, and even assisting with creative writing. I'm known for my ability to generate human-like text based on the input it receives. How can I help you today?
INFO:httpx:HTTP Request: POST http://localhost:11434/api/chat "HTTP/1.1 200 OK"
INFO
I would like you to implement a module named TopModule with the following
interface. All input and output ports are one bit unless otherwise
specified.

      - input  a
      - input  b
      - output out

     The module should implement a NOR gate.

Traceback (most recent call last):
File "/home/yh/my_script/MAGE1/MAGE/tests/test_rtl_generator.py", line 66, in
main()
File "/home/yh/my_script/MAGE1/MAGE/tests/test_rtl_generator.py", line 60, in main
is_pass, code = rtl_gen.chat(spec, key)
^^^^^^^^^^^^^^^^^^^^^^^
TypeError: RTLGenerator.chat() missing 2 required positional arguments: 'interface' and 'rtl_path'
(mage)`

1.1 我尝试填写了以下缺失的值：
is_pass, code = rtl_gen.chat(spec, key, None, "./rtl_out/test.v")
但是依然有别的问题。

2、如果需要实时向ai提问，当前环境下似乎还没实现？我理解当前tests目录下的只是在跑测试用例集合。

[HejiaZ2023]

1. test_rtl_generator是开发过程中的smoke test，确实没有及时更新。我更新了一下，现在在我本地可以跑通了，你checkout到PR的side branch上先试用一下？Fix for rtl generator smoke test #15
2. 是的，这在我们的后续开发计划上。

[ddppt-yy]

似乎还是不行：
首先这些是我修改的代码

diff --git a/pyproject.toml b/pyproject.toml
index 79709d0..1711d9b 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -16,6 +16,9 @@ dependencies = [
   "llama-index-llms-anthropic",
   "llama-index-llms-openai",
   "llama-index-llms-vertex",
+  "llama-index-llms-ollama",
+  "backoff",
+  "pandas",
   "pre-commit",
   "pydantic",
   "rich",

diff --git a/src/mage/gen_config.py b/src/mage/gen_config.py
index 0f69e5c..4343528 100644
--- a/src/mage/gen_config.py
+++ b/src/mage/gen_config.py
@@ -8,6 +8,10 @@ from llama_index.llms.openai import OpenAI
 from llama_index.llms.vertex import Vertex
 from pydantic import BaseModel

+from llama_index.llms.ollama import Ollama
+
+
+
 from .log_utils import get_logger
 from .utils import VertexAnthropicWithCredentials

@@ -103,6 +107,12 @@ def get_llm(**kwargs) -> LLM:

         except Exception as e:
             raise Exception(f"gen_config: Failed to get {provider} LLM") from e
+    elif kwargs["provider"] == "ollama":
+        llm: LLM = Ollama(
+            model=kwargs["model"],
+            request_timeout=60.0,
+        )
+        print( llm.complete("who are you?"))
     else:
         raise ValueError(f"gen_config: Invalid provider: {provider}")

diff --git a/src/mage/token_counter.py b/src/mage/token_counter.py
index 20f186a..eb6f42e 100644
--- a/src/mage/token_counter.py
+++ b/src/mage/token_counter.py
@@ -9,6 +9,7 @@ from llama_index.core.llms.llm import LLM
 from llama_index.llms.anthropic import Anthropic
 from llama_index.llms.openai import OpenAI
 from llama_index.llms.vertex import Vertex
+from llama_index.llms.ollama import Ollama
 from pydantic import BaseModel
 from vertexai.preview.generative_models import GenerativeModel

@@ -142,6 +143,8 @@ class TokenCounter:

             self.encoding = VertexEncoding(llm._client)
             self.activate_structure_output = True
+        elif isinstance(llm, Ollama):
+            pass
         else:
             raise Exception(f"gen_config: No tokenizer for model {model}")
         logger.info(f"Found tokenizer for model '{model}'")

##########################################################
# 以下为 tests内修改的内容
##########################################################
+#     "model": "claude-3-7-sonnet@20250219",
+#     # "model": "gemini-2.0-flash-001",
+#     # "model": "claude-3-7-sonnet-20250219",
+#     # "model": "gpt-4o-2024-08-06",
+#     # "filter_instance": "^(Prob070_ece241_2013_q2|Prob151_review2015_fsm)$",
+#     "filter_instance": "^(Prob011_norgate)$",
+#     # "filter_instance": "^(.*)$",
+#     "type_benchmark": "verilog_eval_v2",
+#     "path_benchmark": "./verilog-eval",
+#     "run_identifier": "your_run_identifier",
+#     "n": 1,
+#     "temperature": 0.85,
+#     "top_p": 0.95,
+#     "max_token": 8192,
+#     "use_golden_tb_in_mage": True,
+#     "key_cfg_path": "./key.cfg",
+# }
+
+
+
+
 args_dict = {
-    "provider": "vertexanthropic",
-    "model": "claude-3-7-sonnet@20250219",
+    "provider": "ollama",
+    "model": "qwen2.5-coder:0.5b",
+    # "model": "claude-3-7-sonnet@20250219",
     # "model": "gemini-2.0-flash-001",
     # "model": "claude-3-7-sonnet-20250219",
     # "model": "gpt-4o-2024-08-06",
@@ -38,6 +62,7 @@ args_dict = {
     "max_token": 8192,
     "use_golden_tb_in_mage": True,
     "key_cfg_path": "./key.cfg",
+    "request_timeout": 60.0,
 }

[ddppt-yy]

以下为当前按照你说的修改后报错：

$ python tests/test_rtl_generator.py
Traceback (most recent call last):
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/httpx/_transports/default.py", line 101, in map_httpcore_exceptions
yield
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/httpx/_transports/default.py", line 250, in handle_request
resp = self._pool.handle_request(req)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/httpcore/_sync/connection_pool.py", line 256, in handle_request
raise exc from None
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/httpcore/_sync/connection_pool.py", line 236, in handle_request
response = connection.handle_request(
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/httpcore/_sync/connection.py", line 103, in handle_request
return self._connection.handle_request(request)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/httpcore/_sync/http11.py", line 136, in handle_request
raise exc
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/httpcore/_sync/http11.py", line 106, in handle_request
) = self._receive_response_headers(**kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/httpcore/_sync/http11.py", line 177, in _receive_response_headers
event = self._receive_event(timeout=timeout)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/httpcore/_sync/http11.py", line 217, in _receive_event
data = self._network_stream.read(
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/httpcore/_backends/sync.py", line 126, in read
with map_exceptions(exc_map):
File "/home/yh/miniconda3/envs/mage/lib/python3.11/contextlib.py", line 158, in exit
self.gen.throw(typ, value, traceback)
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/httpcore/_exceptions.py", line 14, in map_exceptions
raise to_exc(exc) from exc
httpcore.ReadTimeout: timed out

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
File "/home/yh/my_script/MAGE1/MAGE/tests/test_rtl_generator.py", line 95, in
main()
File "/home/yh/my_script/MAGE1/MAGE/tests/test_rtl_generator.py", line 43, in main
llm = get_llm(
^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/mage/gen_config.py", line 115, in get_llm
print( llm.complete("who are you?"))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/llama_index/core/instrumentation/dispatcher.py", line 322, in wrapper
result = func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/llama_index/core/llms/callbacks.py", line 431, in wrapped_llm_predict
f_return_val = f(_self, *args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/llama_index/llms/ollama/base.py", line 511, in complete
return chat_to_completion_decorator(self.chat)(prompt, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/llama_index/core/base/llms/generic_utils.py", line 179, in wrapper
chat_response = func(messages, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/llama_index/core/instrumentation/dispatcher.py", line 322, in wrapper
result = func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/llama_index/core/llms/callbacks.py", line 173, in wrapped_llm_chat
f_return_val = f(_self, messages, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/llama_index/llms/ollama/base.py", line 322, in chat
response = self.client.chat(
^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/ollama/_client.py", line 333, in chat
return self._request(
^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/ollama/_client.py", line 178, in _request
return cls(**self._request_raw(*args, **kwargs).json())
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/ollama/_client.py", line 118, in _request_raw
r = self._client.request(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/httpx/_client.py", line 825, in request
return self.send(request, auth=auth, follow_redirects=follow_redirects)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/httpx/_client.py", line 914, in send
response = self._send_handling_auth(
^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/httpx/_client.py", line 942, in _send_handling_auth
response = self._send_handling_redirects(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/httpx/_client.py", line 979, in _send_handling_redirects
response = self._send_single_request(request)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/httpx/_client.py", line 1014, in _send_single_request
response = transport.handle_request(request)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/httpx/_transports/default.py", line 249, in handle_request
with map_httpcore_exceptions():
File "/home/yh/miniconda3/envs/mage/lib/python3.11/contextlib.py", line 158, in exit
self.gen.throw(typ, value, traceback)
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/httpx/_transports/default.py", line 118, in map_httpcore_exceptions
raise mapped_exc(message) from exc
httpx.ReadTimeout: timed out
(mage)

[HejiaZ2023]

看起来和test_rtl_generator的内容无关，应该是你添加的Ollama部分在get_llm一开始就没有成功；

File "/home/yh/my_script/MAGE1/MAGE/tests/test_rtl_generator.py", line 95, in
main()
File "/home/yh/my_script/MAGE1/MAGE/tests/test_rtl_generator.py", line 43, in main
llm = get_llm(
^^^^^^^^
File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/mage/gen_config.py", line 115, in get_llm
print( llm.complete("who are you?"))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^

建议你:

1. 先在本地把llama-index + Ollama跑通，确认模型配置本身没有问题；
2. 再用你目前的config把test_llm_chat.py调通，确认你修改的MAGE接口代码无误；
3. 然后跑test_rtl_generator, test_top_agent等其他测试。

[ddppt-yy]

好的，因该是模型间歇性不稳定导致的。

还有个问题需要确认下

diff --git a/src/mage/token_counter.py b/src/mage/token_counter.py
index 20f186a..eb6f42e 100644
--- a/src/mage/token_counter.py
+++ b/src/mage/token_counter.py
@@ -9,6 +9,7 @@ from llama_index.core.llms.llm import LLM
 from llama_index.llms.anthropic import Anthropic
 from llama_index.llms.openai import OpenAI
 from llama_index.llms.vertex import Vertex
+from llama_index.llms.ollama import Ollama
 from pydantic import BaseModel
 from vertexai.preview.generative_models import GenerativeModel

@@ -142,6 +143,8 @@ class TokenCounter:

             self.encoding = VertexEncoding(llm._client)
             self.activate_structure_output = True
+        elif isinstance(llm, Ollama):
+            pass
         else:
             raise Exception(f"gen_config: No tokenizer for model {model}")
         logger.info(f"Found tokenizer for model '{model}'")

之前说的token_counter需要修改，这样是否正确？

Traceback (most recent call last):
  File "/home/yh/my_script/MAGE1/MAGE/tests/test_rtl_generator.py", line 95, in <module>
    main()
  File "/home/yh/my_script/MAGE1/MAGE/tests/test_rtl_generator.py", line 81, in main
    is_pass, code = rtl_gen.chat(
                    ^^^^^^^^^^^^^
  File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/mage/rtl_generator.py", line 232, in chat
    response = self.generate(self.history + self.get_order_prompt_messages())
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/mage/rtl_generator.py", line 133, in generate
    resp, token_cnt = self.token_counter.count_chat(messages)
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/mage/token_counter.py", line 174, in count_chat
    in_token_cnt = self.count(llm.messages_to_prompt(messages))
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/yh/miniconda3/envs/mage/lib/python3.11/site-packages/mage/token_counter.py", line 163, in count
    if self.encoding is None:
       ^^^^^^^^^^^^^
AttributeError: 'TokenCounter' object has no attribute 'encoding'
(mage)

我曾尝试解决这个问题 但是没有有效的解决

[HejiaZ2023]

有其他同学向我反馈过类似的问题（添加新模型支持被卡在tokenizer上），看起来确实这里不应该直接crash。
我刚merge了一个新PR把这个raise exception改成了warning，你试用一下？