Configuration of pytorch deep learning environment based on cuda10.0

be based on cuda10.0 Of pytorch Deep learning environment configuration ( Error reporting solution )

1. First look at yourself nvidia Version of the graphics card , Generally applicable cuda10.0：

1. open win+s Search for nvidia Control panel , Check system information

2. You can see own CUDA Version number of 496.76,

3. In the installation CUDA Before , You need to check your computer GPU Driver version , because CUDA Version pair GPU The driver version has requirements , The correspondence is as follows ：

Therefore, it fully meets the installation cuda10.0

2. install cuda10.0 and cudnn

Install Links ：

CUDA and cuDNN install 10.0 edition

3. Create and install pytorch Environmental Science ：

1、 Corresponding cuda10.0 Of pytorch Installation command ：

Installed pytorch Version needs and cuda Versions, ,orch Official website

Looking for historical versions ：

The installation command is as follows ：

# CUDA 10.0
conda install pytorch==1.2.0 torchvision==0.4.0 cudatoolkit=10.0 -c pytorch

2、 establish pytorch_gpu Environmental Science ：

Get into Anaconda prompt The command interface is created with the name pytorch Virtual environment for ：

# python The version number is 3.7：

conda create -n pytorch python=3.7

Use activate Activate the environment

conda activate pytorch

Enter... Above pytorch Installation command ：

If there is an installation error , Show pytorch The website needs to be downloaded manually torch package ：

Don't rush out of the command line interface ！！！！！！, Download it torch Continue operation after package .

link ： Baidu SkyDrive torch package
Extraction code ：2022

Copy to Anaconda3 Under the designated path ：C:\Users\X\anaconda3\pkgs

Then continue typing in the command line interface ：

conda install --use-local pytorch-1.7.0-py3.7_cuda101_cudnn7_0.tar.bz2

You can install it successfully .

4. Verify that the environment is installed successfully , Can successfully call gpu：

stay Anaconda prompt Command interface input activate pytorch, Activate the environment ,

Input python

introduce torch

import torch

torch.cuda.is_available（）

If it shows True Indicates successful installation .