Download: 1.13.1 https://download.pytorch.org/whl/cu116/torch-1.13.1%2Bcu116-cp39-cp39-win_amd64.whl https://download.pytorch.org/whl/cu116/torchvision-0.14.1%2Bcu116-cp39-cp39-win_amd64.whl https://download.pytorch.org/whl/cu116/torchaudio-0.13.1%2Bcu116-cp39-cp39-win_amd64.whl 下载安装CUDA 11.6.2 https://developer.nvidia.com/cuda-toolkit-archive 下载cudnn8.9.2: https://developer.nvidia.com/rdp/cudnn-archive cudnn-windows-x86_64-8.9.2.26_cuda11-archive.zip 解压到CUDA安装目录，如C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6 pip install xxx.whl 测试 import torch print(torch.cuda.is_available()) print(torch.__version__) 输出 True 1.13.1+cu116 则成功 修改PyTorch预训练模型的下载路径 import os os.environ['TORCH_HOME']='D:/pytorch-model/' 旧版1.11.0: https://download.pytorch.org/whl/cu113/torch-1.11.0%2Bcu113-cp37-cp37m-win_amd64.whl https://download.pytorch.org/whl/cu113/torchvision-0.12.0%2Bcu113-cp37-cp37m-win_amd64.whl https://download.pytorch.org/whl/cu113/torchaudio-0.11.0%2Bcu113-cp37-cp37m-win_amd64.whl