Anaconda Pytorch windows Environment Settings

原创已于 2022-08-14 16:10:46 修改 · 508 阅读

0 ·

本内容遵循CC 4.0 BY-SA版权协议

标签

#pytorch #windows #python

收录于

于 2022-08-14 16:03:33 首次发布

Computer Vision 专栏收录该内容

2 篇文章

订阅专栏

本文介绍如何在Windows环境下安装最新版Anaconda，并设置兼容Python 3.9的虚拟环境。此外，还详细说明了如何根据GPU计算能力和CUDA版本选择合适的PyTorch版本，确保深度学习项目的顺利运行。

Anaconda

Url：https://www.anaconda.com
Download the newest version(2022.05.windows.x86-64) of the anaconda for windows, the highest-level comparability reach to python 3.9(64-bit)
在这里插入图片描述

CUDA

wiki: search CUDA
Figure out GPU’s compute capability-8.6
在这里插入图片描述
Check the CUDA Driver Version

Choose the proper CUDA Runtime Version≤ CUDA Driver Version

PyTorch, choose the right CUDA Runtime Version = 11.6，typing following command in virtual environment(followings are two choices)

Failed this method

Successfully working!

Test it out

import torch
torch.cuda.is_available() #output : true

Pycharm settings

在这里插入图片描述

commands for conda

conda env list  #find which environments are installed
conda create -n <virtual environment name> python=<version> [-c https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main] #create env using channel or not
conda activate <virtual environment name> # entry the env
conda deactivate  # exit virtual environment
conda remove -n <virtual environment name> --all # remove virtual environment
conda uninstall <pkgname> # pip is often used and recommended
conda install <pkgname>