华为计算微信公众号
昇腾AI开发者公众号
华为计算微博
华为计算今日头条
型号:Atlas 300i Duo
已经在conda虚拟环境中安装好torch_npu,但启动MindIE Server后日志报错,尝试vLLM文本推理接口,发送请求也失败。
日志详细报错信息如下:
2024-10-18 07:22:13.542084 1558116 Log started at [trace] level 2024-10-18 07:22:13.542125 1558116 Log default format: yyyy-mm-dd hh:mm:ss.uuuuuu threadid loglevel msg 2024-10-18 07:22:13.542087 1558117 Log started at [trace] level 2024-10-18 07:22:13.542128 1558117 Log default format: yyyy-mm-dd hh:mm:ss.uuuuuu threadid loglevel msg 2024-10-18 07:22:13.542718 1558118 Log started at [trace] level 2024-10-18 07:22:13.542759 1558118 Log default format: yyyy-mm-dd hh:mm:ss.uuuuuu threadid loglevel msg 2024-10-18 07:22:13.542718 1558119 Log started at [trace] level 2024-10-18 07:22:13.542763 1558119 Log default format: yyyy-mm-dd hh:mm:ss.uuuuuu threadid loglevel msg 2024-10-18 07:22:14.782506 1558124 error LLMModelExecutorPython.cpp:64] [Connector]class LLMPythonModel initializes fail.N8pybind1117error_already_setE, ModuleNotFoundError: No module named 'torch_npu' At: /home/HwHiAiUser/miniconda3/envs/llm/lib/python3.10/site-packages/model_wrapper/standard_model.py(8): <module> <frozen importlib._bootstrap>(241): _call_with_frames_removed <frozen importlib._bootstrap_external>(883): exec_module <frozen importlib._bootstrap>(703): _load_unlocked <frozen importlib._bootstrap>(1006): _find_and_load_unlocked <frozen importlib._bootstrap>(1027): _find_and_load /home/HwHiAiUser/miniconda3/envs/llm/lib/python3.10/site-packages/model_wrapper/model.py(8): <module> <frozen importlib._bootstrap>(241): _call_with_frames_removed <frozen importlib._bootstrap_external>(883): exec_module <frozen importlib._bootstrap>(703): _load_unlocked <frozen importlib._bootstrap>(1006): _find_and_load_unlocked <frozen importlib._bootstrap>(1027): _find_and_load /home/HwHiAiUser/miniconda3/envs/llm/lib/python3.10/site-packages/model_wrapper/__init__.py(5): <module> <frozen importlib._bootstrap>(241): _call_with_frames_removed <frozen importlib._bootstrap_external>(883): exec_module <frozen importlib._bootstrap>(703): _load_unlocked <frozen importlib._bootstrap>(1006): _find_and_load_unlocked <frozen importlib._bootstrap>(1027): _find_and_load 2024-10-18 07:22:14.782605 1558124 error slave_IPC_communicator.cpp:317] [SlaveIPCCommunicator::ProcessBroadcastRecvMessage]Executor failed to init model 2024-10-18 07:22:14.782631 1558124 error slave_IPC_communicator.cpp:438] [HandleRcvMsg]Executor failed to process recv broadcastMessage 2024-10-18 07:22:14.782640 1558124 error slave_IPC_communicator.cpp:446] [HandleRcvMsg]slave communicator notify connector to terminate 2024-10-18 07:22:14.782649 1558124 warning main.cpp:24] [Backend connector]Signal 15 received, exiting the process 2024-10-18 07:22:14.815525 1558130 error LLMModelExecutorPython.cpp:64] [Connector]class LLMPythonModel initializes fail.N8pybind1117error_already_setE, ModuleNotFoundError: No module named 'torch_npu'
我要发帖子
型号:Atlas 300i Duo
已经在conda虚拟环境中安装好torch_npu,但启动MindIE Server后日志报错,尝试vLLM文本推理接口,发送请求也失败。
日志详细报错信息如下: