Win10 uses Tsinghua source to quickly install pytorch-GPU version (recommended)

Check whether your cuda is installed

Type in the anaconda prompt

nvcc -V

The display above indicates that the installation is complete.

Configure Tsinghua Park download environment

Also enter in the anaconda prompt

conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/msys2/ 
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge/ 
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/ 
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/peterjc123/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/
conda config --set show_channel_urls yes

Then delete the .condarc file in the C:\Users\Administrator directory, open the file and delete the defaults line. As shown in the following figure:

Click to enter the pytorch official website

Official website link: https://pytorch.org/get-started/locally/.

Choose your own version, as follows:

conda install pytorch torchvision cudatoolkit=10.1 -c pytorch

The meaning of this line of code is: when installing pytorch, torchvision will also be installed, so there is no need to install torchvision later. cudatoolkit=10.1 means cuda version 10.1. -c means downloading from the official website. Without -cpu, it indicates GPU version.

Configure your own cuda version of pytorch, what to do if the cuda version is too low

As mentioned above, just change cudatoolkit=10.1 to your own cuda version, such as mine is cuda 9.0:

conda install pytorch torchvision cudatoolkit=9.0 pytorch

Just install it.

Install the specified version of torchvision

When you are looking at the code, sometimes the author requires a specific version of torch or torchvision. This can only be done by downloading previous versions of the file on the official website.

Click Previous PyTorch Versions

The corresponding file of the version you want to find is above, just download it and pip install it. There are many version issues that need to be noted, such as: torchvison of cuda9.0 only supports version 0.3.0, etc.

Verify that your pytorch is installed

That's it!

Python
import torch
exit()

Congratulations on your successful installation.

Summarize

The above is the editor's introduction to win10 using Tsinghua source to quickly install pytorch-GPU version. I hope it will be helpful to everyone. If you have any questions, please leave me a message and the editor will reply to you in time. I would also like to thank everyone for their support of the 123WORDPRESS.COM website!
If you find this article helpful, please feel free to reprint it and please indicate the source. Thank you!