update python env

Author: shell-nlp

docker-compose.yml

@@ -10,7 +10,7 @@ services:
     # image: docker.1ms.run/506610466/gpt_server:latest
     image: gpt_server:latest_
     container_name: gpt_server
-    shm_size: '8g' # 设置共享内存为4GB 
+    shm_size: '32g' # 设置共享内存为4GB 
     restart: always
     # network_mode: host
     ports:

gpt_server/model_backend/lmdeploy_backend.py

@@ -25,10 +25,11 @@
     "lmdeploy-pytorch": "pytorch",  # pytorch后端
     "lmdeploy-turbomind": "turbomind",  # turbomind后端
 }
+log_level = os.getenv("log_level", "WARNING")
 from lmdeploy.utils import get_logger
 
-get_logger("lmdeploy").setLevel("WARNING")
-os.environ["TM_LOG_LEVEL"] = "ERROR"
+get_logger("lmdeploy").setLevel(log_level)  # 默认WARNING
+os.environ["TM_LOG_LEVEL"] = "WARNING"
 
 
 def is_stop(output: str, stop_str: str):

gpt_server/serving/openai_api_server.py

@@ -308,7 +308,6 @@ def get_gen_params(
     # ------- TODO add messages tools -------
     gen_params["response_format"] = response_format
     gen_params["reasoning_parser"] = reasoning_parser
-    logger.debug(f"==== request ====\n{gen_params}")
     return gen_params
 
 

pyproject.toml

@@ -12,14 +12,14 @@ dependencies = [
     "ffmpy",
     "fschat==0.2.36",
     "infinity-emb[all]==0.0.76",
-    "lmdeploy==0.8.0",
+    "lmdeploy==0.9.0",
     "loguru>=0.7.2",
     "openai==1.86.0",
     "setuptools==75.2.0",
     "streamlit==1.39.0",
     "torch==2.6.0",
     "torchvision==0.20.1",
-    "vllm==0.9.0.1",
+    "vllm==0.9.1",
     "qwen_vl_utils",
     "evalscope[perf,rag]==0.16.1",
     "modelscope==1.26.0",
@@ -35,7 +35,7 @@ dependencies = [
 default-groups = [] # 默认只安装dependencies中的库
 override-dependencies = [
     "setuptools==75.2.0",
-    "torchvision==0.22.0",
+    "torchvision==0.22.1",
     "torch==2.7.0",
     "triton",
     "outlines==0.1.11",

requirements.txt

@@ -27,7 +27,7 @@ aiohttp==3.12.13
     #   vllm
 aiosignal==1.3.2
     # via aiohttp
-airportsdata==20250523
+airportsdata==20250622
     # via outlines
 aliyun-python-sdk-core==2.16.0
     # via
@@ -39,7 +39,7 @@ altair==5.5.0
     # via streamlit
 annotated-types==0.7.0
     # via pydantic
-anthropic==0.54.0
+anthropic==0.55.0
     # via sglang
 antlr4-python3-runtime==4.9.3
     # via
@@ -122,7 +122,7 @@ coloredlogs==15.0.1
     # via onnxruntime
 colpali-engine==0.3.10
     # via infinity-emb
-compressed-tensors==0.9.4
+compressed-tensors==0.10.1
     # via
     #   sglang
     #   vllm
@@ -160,7 +160,7 @@ decord==0.6.0
     # via sglang
 depyf==0.18.0
     # via vllm
-diffusers==0.33.1
+diffusers==0.34.0
     # via gpt-server (pyproject.toml)
 dill==0.3.8
     # via
@@ -298,7 +298,7 @@ httpx==0.28.1
     #   langsmith
     #   litellm
     #   openai
-httpx-sse==0.4.0
+httpx-sse==0.4.1
     # via langchain-community
 huggingface-hub==0.33.0
     # via
@@ -412,7 +412,7 @@ langchain-core==0.3.66
     #   langchain-openai
     #   langchain-text-splitters
     #   ragas
-langchain-openai==0.3.24
+langchain-openai==0.3.25
     # via
     #   evalscope
     #   ragas
@@ -439,9 +439,9 @@ librosa==0.11.0
     # via
     #   flashtts
     #   funasr
-litellm==1.72.7
+litellm==1.73.1
     # via sglang
-llguidance==0.7.29
+llguidance==0.7.30
     # via
     #   sglang
     #   vllm
@@ -451,7 +451,7 @@ llvmlite==0.44.0
     #   pynndescent
 lm-format-enforcer==0.10.11
     # via vllm
-lmdeploy==0.8.0
+lmdeploy==0.9.0
     # via gpt-server (pyproject.toml)
 loguru==0.7.3
     # via gpt-server (pyproject.toml)
@@ -499,15 +499,15 @@ msgspec==0.19.0
     #   vllm
 mteb==1.38.20
     # via evalscope
-multidict==6.5.0
+multidict==6.5.1
     # via
     #   aiohttp
     #   yarl
 multiprocess==0.70.16
     # via datasets
 mypy-extensions==1.1.0
     # via typing-inspect
-narwhals==1.43.1
+narwhals==1.44.0
     # via altair
 nest-asyncio==1.6.0
     # via
@@ -737,7 +737,7 @@ pandas==2.3.0
     #   streamlit
 parso==0.8.4
     # via jedi
-partial-json-parser==0.2.1.1.post5
+partial-json-parser==0.2.1.1.post6
     # via
     #   lmdeploy
     #   sglang
@@ -854,11 +854,11 @@ pydantic==2.11.7
     #   xgrammar
 pydantic-core==2.33.2
     # via pydantic
-pydantic-settings==2.9.1
+pydantic-settings==2.10.1
     # via langchain-community
 pydeck==0.9.1
     # via streamlit
-pygments==2.19.1
+pygments==2.19.2
     # via
     #   ipython
     #   markdown2
@@ -881,7 +881,7 @@ python-dateutil==2.9.0.post0
     #   modelscope
     #   pandas
     #   posthog
-python-dotenv==1.1.0
+python-dotenv==1.1.1
     # via
     #   litellm
     #   pydantic-settings
@@ -1153,7 +1153,7 @@ tiktoken==0.9.0
     #   xgrammar
 timm==1.0.15
     # via infinity-emb
-tokenizers==0.21.1
+tokenizers==0.21.2
     # via
     #   litellm
     #   transformers
@@ -1195,7 +1195,7 @@ torchao==0.9.0
     # via sglang
 torchaudio==2.7.0
     # via vllm
-torchvision==0.22.0
+torchvision==0.22.1
     # via
     #   --override (workspace)
     #   gpt-server (pyproject.toml)
@@ -1321,7 +1321,7 @@ uvloop==0.21.0
     # via
     #   sglang
     #   uvicorn
-vllm==0.9.0.1
+vllm==0.9.1
     # via gpt-server (pyproject.toml)
 watchdog==5.0.3
     # via streamlit