Shulou(Shulou.com)06/01 Report--

This article mainly explains "how to install Python+CUDA+PyTorch under Window". The content in the article is simple and clear, and it is easy to learn and understand. Please follow the editor's train of thought to study and learn how to install Python+CUDA+PyTorch under Window.

1 Overview

Python+CUDA+PyTorch installation under Windows, the steps are very detailed, hereby recorded to help readers avoid detours.

2 Python

The installation of Python is relatively simple. Download the exe installation package from the official website:

Because the latest version of torch only supports Python 3.8, it is recommended to install Python 3.8 in order to avoid version compatibility problems.

Download the middle executable installer and install it. If you open it, it is recommended to choose customization:

Select the required components:

Select a location:

After installation, you can check whether the installation is successful from the command line (generally, the path will be added to the environment variable by default):

Python-version

Since the installation package automatically installs pip, you will be prompted that pip needs to be upgraded when you use it. Use the following command to upgrade pip:

Python-m pip install-- upgrade pip

It is recommended to add a domestic source before installation, create a new pip folder under the user directory, and create a new file called pip.ini in it. Enter the following:

[global] index-url = https://pypi.tuna.tsinghua.edu.cn/simple[install]trusted-host = https://pypi.tuna.tsinghua.edu.cn3 CUDA

It is not difficult to install CUDA. Generally speaking, Windows automatically installs the corresponding N card driver when installing the system, and downloads the corresponding CUDA Toolkit from the official website:

Open it after downloading (it is actually a compressed package, here is the decompression path):

The installation interface will be opened automatically after the decompression is completed:

If there is no special need, you can use the default installation option directly, otherwise, select a custom component to install:

Since the author has not installed VS, you will be prompted that VS cannot be found. Just ignore it:

The next step is to wait for the installation:

Installation completed:

Open cmd and enter nvcc-- version. The installation is successful if the version number is displayed:

4 PyTorch4.1 download

PyTorch is recommended to install manually offline. Download the torch and torchvision of the corresponding CUDA version and the corresponding Python version here:

The first cu110 represents cuda 11.0, then the torch/torchvision, then the version, and then the cp represents the Python version, such as cp36 for Python 3.6, followed by the system and architecture.

4.2 installation

After the download is complete, you can install it using pip:

Pip install torch-1.7.0+cu110-cp38-cp38-win_amd64.whlpip install torchvision-0.8.1+cu110-cp38-cp38-win_amd64.whl

Note that the following problems may occur in the Windows 2004 version:

This is a problem with numpy version 19.0.4. For more information, please click here.

The workaround is to install version 19.0.3:

Pip uninstall numpypip install numpy==19.0.3

Then test and open python:

Pythonimport torchprint (torch.__version__)

The output of the torch version indicates success:

Thank you for reading, the above is the content of "how to install Python+CUDA+PyTorch under Window". After the study of this article, I believe you have a deeper understanding of how to install Python+CUDA+PyTorch under Window, and the specific use needs to be verified in practice. Here is, the editor will push for you more related knowledge points of the article, welcome to follow!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.