(Chinese_Vicuna_master) C:\Windows\System32\Chinese-Vicuna-master>python finetune.py --data_path merge.json --test_size 2000
Traceback (most recent call last):
File "finetune.py", line 4, in
import torch
File "C:\Users\XJM2022\anaconda3\envs\Chinese_Vicuna_master\lib\site-packages\torch__init__.py", line 139, in
raise err
OSError: [WinError 126] 找不到指定的模块。 Error loading "C:\Users\XJM2022\anaconda3\envs\Chinese_Vicuna_master\lib\site-packages\torch\lib\shm.dll" or one of its dependencies.
(Chinese_Vicuna_master) C:\Windows\System32\Chinese-Vicuna-master>pip list
Package Version
certifi 2022.12.7
pip 23.0.1
setuptools 65.6.3
torch 1.13.1
wheel 0.38.4
wincertstore 0.2
WARNING: The repository located at pypi.tuna.tsinghua.edu.cn is not a trusted or secure host and is being ignored. If this repository is available via HTTPS we recommend you use HTTPS instead, otherwise you may silence this warning and allow it anyway with '--trusted-host pypi.tuna.tsinghua.edu.cn'.
我看到说明里面讲要用cuda12的版本?但好像pytorch官网上面只有对应cuda11.x的版本,我下载的是pytorch-1.13.1-py3.8_cuda11.7_cudnn8_0. 不知道是否正确
我使用conda的虚拟环境部署,由于在线加入torch速度太慢,就用了离线的方式下载了pytorch-1.13.1-py3.8_cuda11.7_cudnn8_0.tar.bz2 ,然后用离线安装的方式加载了
conda install --offline C:\AI\pytorch\pytorch-1.13.1-py3.8_cuda11.7_cudnn8_0.tar.bz2
但到了执行 python finetune.py --data_path merge.json --test_size 2000 这一步的时候,报错提示这是什么原因呢?
附:我在conda list里面看到的是
pip list看到的是
感谢。