How to install PyTorch under Linux

1. Introduction to PyTorch

PyTorch is an open source Python machine learning library based on Torch and used for natural language processing and other applications. In January 2017, PyTorch was launched based on Torch by the Facebook Artificial Intelligence Research Institute (FAIR). The predecessor of PyTorch is Torch. Its underlying layer is the same as the Torch framework, but a lot of content has been rewritten in Python. It is not only more flexible, supports dynamic graphics, but also provides a Python interface. Developed by the Torch7 team, it is a Python-first deep learning framework that not only enables powerful GPU acceleration, but also supports dynamic neural networks. PyTorch can be regarded as numpy with GPU support, and it can also be regarded as a powerful deep neural network with automatic derivation function. In addition to Facebook, it has been adopted by institutions such as Twitter, CMU and Salesforce.

2. Installation steps

1. Operating system selection

  View PyTorch official website, you can see that PyTorch supports Linux, Mac, window platform, supports conda, Installation methods such as pip and source code also support CPU, cuda, and ROCm computing platforms. When we click on the environment selection, we can find that currently only the Linux system supports all languages, all installation methods, and all computing platforms, so we choose the Linux operating system as the system environment. . In addition, machine learning calculations require a higher version of glibc, the kernel and glbic versions of centos are lower, and the kernel versions of Ubuntu are newer, so it is recommended to use the Ubuntu operating system for machine learning hosts. Currently, the minimum Ubuntu version supported by cuda update is 18.04, so it is recommended to use an operating system of Ubuntu 18.04 or above.

wuhs@s169:~$ cat /etc/os-release
NAME="Ubuntu"
VERSION="18.04.6 LTS ( Bionic Beaver)"

2. Anaconda3 installation

  As shown above, PyTorch supports a variety of installation methods. The blogger plans to use the conda installation method. It is recommended to first After installing Anaconda3, we can create different virtual environments according to our needs. Different PyTorch versions can be installed in the virtual environment. The virtual environments support each other without affecting each other. For the installation of anaconda in Ubuntu environment, see the blog post Ubuntu Anaconda3 Installation.

wuhs@s169:~$ wget https://mirrors.bfsu.edu.cn/anaconda/archive/Anaconda3-2022.10-Linux-x86_64.sh
wuhs@s169:~$ sh Anaconda3-2022.10-Linux-x86_64.sh
wuhs@s169:~$ source ~/.bashrc

#3. Check the Python version

  different The Python versions required by PyTorch versions are different, so after installing anaconda3, we check the current Python version. The default initialization is the latest version of Python corresponding to the current anaconda3 release. Of course, we can also use conda to create the required Python environment version. . We check PyTorch, torchvision, and Python version matching requirements at torchvision.

(base) wuhs@s169:~$ python -V
Python 3.9.13

4. Install PyTorch

  As shown in the second step, on the PyTorch official website, we can generate the corresponding installation command after selecting the operating system, installation method, programming language, and computing platform.

(base) wuhs@s169:~$ conda install pytorch torchvision torchaudio cpuonly -c pytorch
Collecting package metadata (current_repodata.json): done
Solving environment: done

## Package Plan

##Proceed ([y]/n)? y
…

5. Version verification

(base) wuhs@s169:~$ python
Python 3.9.13 (main, Aug 25 2022, 23:26:10)
[GCC 11.2 .0] :: Anaconda, Inc. on linux
Type “help”, “copyright”, “credits” or “license” for more information.
>>> import torch
> >> torch.version
‘1.13.1’
>>>

3. Installation of specified version

1. Create a virtual environment

(base) wuhs@s169:~$ conda create -n pytorch python=3.9
…
(base) wuhs@s169:~$ conda activate pytorch
(pytorch) wuhs@s169:~$

2. Install the specified version of PyTorch

  When installing the specified version of PyTorch, we need to Check the matching version on the PyTorch channel on the GitHub official website. Specify the version number when installing conda. For the specific version number, you can check the anaconda official website. For the corresponding relationship between the PyTorch version and TorchAudio, see
TorchAudio. Of course, if we specify the wrong version, an error will be reported during installation. We can check which software version is wrong according to the error message, and then go to the official website to confirm the correction and reinstall.

(pytorch) wuhs@s169:~$ conda install pytorch2.12.0 torchvision=0.13.0 torchaudio0.12.0 cpuonly -c pytorch

The above is the detailed content of How to install PyTorch under Linux. For more information, please follow other related articles on the PHP Chinese website!