docker 部署 MindIE 仍然缺少 atb_llm 模块
收藏回复举报
docker 部署 MindIE 仍然缺少 atb_llm 模块
t('forum.solved') 已解决
发表于2024-07-13 19:23:40
0 查看

宿主机环境:NPU驱动版本:23.0.3 NPU固件版本:7.1.0.5.220

docker 环境:

拉取了最基础的ubuntu22.04镜像

启动容器命令:

docker run -it \ --name mindie \ --ipc=host \ --device=/dev/davinci0 \ --device=/dev/davinci1 \ --device=/dev/davinci2 \ --device=/dev/davinci3 \ --device=/dev/davinci_manager \ --device=/dev/devmm_svm \ --device=/dev/hisi_hdc \ -v /usr/local/dcmi:/usr/local/dcmi \ -v /usr/local/bin/npu-smi:/usr/local/bin/npu-smi \ -v /usr/local/Ascend/driver/lib64/common:/usr/local/Ascend/driver/lib64/common \ -v /usr/local/Ascend/driver/lib64/driver:/usr/local/Ascend/driver/lib64/driver \ -v /etc/ascend_install.info:/etc/ascend_install.info \ -v /etc/vnpu.cfg:/etc/vnpu.cfg \ -v /usr/local/Ascend/driver/version.info:/usr/local/Ascend/driver/version.info \ -v /data1/model:/data ubuntu:22.04

docker内部环境:

python --version

Python 3.10.0

pip --version

pip 24.1.1 from /root/.local/lib/python3.10/site-packages/pip (python 3.10)

npu-smi info  可显示当前npu设备信息;

atc --help 输出正常

安装的CANN版本为8.0.RC1版本

---------------------------------------------------------------------------------------------------------------------------

具体的报错信息为:

Use user defined log path: /usr/local/Ascend/mindie/1.0.RC1/mindie-service/logs/mindservice.log 

[Model]    >>> Exception:No module named 'atb_llm' 

Traceback (most recent call last): 

  File "/usr/local/Ascend/mindie/latest/mindie-service/bin/model.py", line 28, in initialize 

    return self.python_model.initialize(config) 

  File "/usr/local/Ascend/mindie/latest/mindie-service/bin/standard_model.py", line 24, in initialize 

    self.model = WrapperFactory.get_model_wrapper(model_config) 

  File "/usr/local/Ascend/mindie/latest/mindie-service/bin/utils/wrapper_factory.py", line 22, in get_model_wrapper 

    from utils.atb_wrapper import StandardATBModelWrapper 

  File "/usr/local/Ascend/mindie/latest/mindie-service/bin/utils/atb_wrapper.py", line 8, in <module> 

    from atb_llm.runner import ModelRunner 

ModuleNotFoundError: No module named 'atb_llm' 

[Model]    >>> return initialize error result: {'status': 'error', 'npuBlockNum': '0', 'cpuBlockNum': '0'} 

[Model]    >>> Exception:No module named 'atb_llm' 

Traceback (most recent call last): 

  File "/usr/local/Ascend/mindie/latest/mindie-service/bin/model.py", line 28, in initialize 

    return self.python_model.initialize(config) 

  File "/usr/local/Ascend/mindie/latest/mindie-service/bin/standard_model.py", line 24, in initialize 

    self.model = WrapperFactory.get_model_wrapper(model_config) 

  File "/usr/local/Ascend/mindie/latest/mindie-service/bin/utils/wrapper_factory.py", line 22, in get_model_wrapper 

    from utils.atb_wrapper import StandardATBModelWrapper 

  File "/usr/local/Ascend/mindie/latest/mindie-service/bin/utils/atb_wrapper.py", line 8, in <module> 

    from atb_llm.runner import ModelRunner 

ModuleNotFoundError: No module named 'atb_llm' 

[Model]    >>> return initialize error result: {'status': 'error', 'npuBlockNum': '0', 'cpuBlockNum': '0'} 

free(): invalid pointer 

free(): invalid pointer 

Daemon is killing... 

Killed 

-----------------------------------------------------------------------------------------------

根据链接docker 部署 MindIE 缺少 atb_llm 模块_昇腾主版块_昇腾论坛 (hiascend.com)

安装最新包

wget -O  "atb_llm-0.0.1-py3-none-any.whl" https://tmp-hd106.vx-cdn.com/file-667e7b9778b0a-667e7bab5e911/atb_llm-0.0.1-py3-none-any.whl

仍然失败

image.png

目前有什么解决办法吗??

本帖最后由 匿名用户2024/07/13 19:24:47 编辑

我要发帖子