Ubuntu20.04 installs Pytorch1.7.0 GPU version torchvision=0.8.1 (CUDA and CUDNN have been installed)

• Create a virtual environment
  conda create -n pytorch python=3.8
conda env list
conda activate pytorch

• Go to the official the corresponding installation command:
  pip install torch==1.7.0+cu101 torchvision==0.8.1+cu101 torchaudio==0.7.0 -f https://download.pytorch.org/whl/torch_stable.html
  
  
  However, the download is a bit slow. I first downloaded Torch with Motrix : https://download.pytorch.org/whl/cu101/torch-1.7.0%2Bcu101-cp38-cp38-linux_x86_64 .whl
  and then install torch first (you need to switch to the directory where you downloaded torch): pip install torch-1.7.0+cu101-cp38-cp38-linux_x86_64.whl -i https://pypi.tuna.tsinghua.edu.cn/simple/
  Finally, use this command to download the rest of the library:pip install torch==1.7.0+cu101 torchvision==0.8.1+cu101 torchaudio==0.7.0 -f https://download.pytorch.org/whl/torch_stable.html

• Test whether the installation is complete:

  import torch
  import torchvision
  
  if __name__ == "__main__":
      print(torch.__version__)  # 查看torch当前版本号
      print(torchvision.__version__)
  
      # 查看CUDA版本号：
      print(torch.version.cuda)  # 编译当前版本的torch使用的cuda版本号
  
      # 测试Cudnn是否安装成功
      from torch.backends import cudnn
      print(cudnn.is_available())
  
      # 是否GPU可用
      print(torch.cuda.is_available())  # 查看当前cuda是否可用于当前版本的Torch，如果输出True，则表示可用