安裝pytorch命令

網址：https://pytorch.org/get-started/previous-versions/

INSTALLING PREVIOUS VERSIONS OF PYTORCH

We’d prefer you install the latest version, but old binaries and installation instructions are provided below for your convenience.

# 配置國內源，安裝PyTorch用
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/

COMMANDS FOR VERSIONS >= 1.0.0

v1.2.0

Conda

OSX

# conda
conda install pytorch==1.2.0 torchvision==0.4.0 -c pytorch

Linux and Windows

# CUDA 9.2
conda install pytorch==1.2.0 torchvision==0.4.0 cudatoolkit=9.2 -c pytorch

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

# CPU Only
conda install pytorch==1.2.0 torchvision==0.4.0 cpuonly -c pytorch

Wheel

OSX

pip install torch==1.2.0 torchvision==0.4.0

Linux and Windows

# CUDA 10.0
pip install torch==1.2.0 torchvision==0.4.0 -f https://download.pytorch.org/whl/torch_stable.html

# CUDA 9.2
pip install torch==1.2.0+cu92 torchvision==0.4.0+cu92 -f https://download.pytorch.org/whl/torch_stable.html

# CPU only
pip install torch==1.2.0+cpu torchvision==0.4.0+cpu -f https://download.pytorch.org/whl/torch_stable.html

v1.1.0

Conda

OSX

# conda
conda install pytorch==1.1.0 torchvision==0.3.0 -c pytorch

Linux and Windows

# CUDA 9.0
conda install pytorch==1.1.0 torchvision==0.3.0 cudatoolkit=9.0 -c pytorch

# CUDA 10.0
conda install pytorch==1.1.0 torchvision==0.3.0 cudatoolkit=10.0 -c pytorch

# CPU Only
conda install pytorch-cpu==1.1.0 torchvision-cpu==0.3.0 cpuonly -c pytorch

Wheel

OSX

pip install torch==1.1.0 torchvision==0.3.0

Linux and Windows

# CUDA 10.0
Download and install wheel from https://download.pytorch.org/whl/cu100/torch_stable.html

# CUDA 9.0
Download and install wheel from https://download.pytorch.org/whl/cu90/torch_stable.html

# CPU only
Download and install wheel from https://download.pytorch.org/whl/cpu/torch_stable.html

v1.0.1

Conda

OSX

# conda
conda install pytorch==1.0.1 torchvision==0.2.2 -c pytorch

Linux and Windows

# CUDA 9.0
conda install pytorch==1.0.1 torchvision==0.2.2 cudatoolkit=9.0 -c pytorch

# CUDA 10.0
conda install pytorch==1.0.1 torchvision==0.2.2 cudatoolkit=10.0 -c pytorch

# CPU Only
conda install pytorch-cpu==1.0.1 torchvision-cpu==0.2.2 cpuonly -c pytorch

Wheel

OSX

pip install torch==1.0.1 torchvision==0.2.2

Linux and Windows

# CUDA 10.0
Download and install wheel from https://download.pytorch.org/whl/cu100/torch_stable.html

# CUDA 9.0
Download and install wheel from https://download.pytorch.org/whl/cu90/torch_stable.html

# CPU only
Download and install wheel from https://download.pytorch.org/whl/cpu/torch_stable.html

v1.0.0

Conda

OSX

# conda
conda install pytorch==1.0.0 torchvision==0.2.1 -c pytorch

Linux and Windows

# CUDA 10.0
conda install pytorch==1.0.0 torchvision==0.2.1 cuda100 -c pytorch

# CUDA 9.0
conda install pytorch==1.0.0 torchvision==0.2.1 cuda90 -c pytorch

# CUDA 8.0
conda install pytorch==1.0.0 torchvision==0.2.1 cuda80 -c pytorch

# CPU Only
conda install pytorch-cpu==1.0.0 torchvision-cpu==0.2.1 cpuonly -c pytorch

Wheel

OSX

pip install torch==1.0.0 torchvision==0.2.1

Linux and Windows

# CUDA 10.0
Download and install wheel from https://download.pytorch.org/whl/cu100/torch_stable.html

# CUDA 9.0
Download and install wheel from https://download.pytorch.org/whl/cu90/torch_stable.html

# CUDA 8.0
Download and install wheel from https://download.pytorch.org/whl/cu80/torch_stable.html

# CPU only
Download and install wheel from https://download.pytorch.org/whl/cpu/torch_stable.html

COMMANDS FOR VERSIONS < 1.0.0

Via conda

This should be used for most previous macOS version installs.

To install a previous version of PyTorch via Anaconda or Miniconda, replace “0.4.1” in the following commands with the desired version (i.e., “0.2.0”).

Installing with CUDA 9

conda install pytorch=0.4.1 cuda90 -c pytorch

or

conda install pytorch=0.4.1 cuda92 -c pytorch

Installing with CUDA 8

conda install pytorch=0.4.1 cuda80 -c pytorch

Installing with CUDA 7.5

conda install pytorch=0.4.1 cuda75 -c pytorch

Installing without CUDA

conda install pytorch=0.4.1 -c pytorch

From source

It is possible to checkout an older version of PyTorch and build it. You can list tags in PyTorch git repository with git tag and checkout a particular one (replace ‘0.1.9’ with the desired version) with

git checkout v0.1.9

Follow the install from source instructions in the README.md of the PyTorch checkout.

Via pip

Download the whl file with the desired version from the following html pages:

• https://download.pytorch.org/whl/cpu/torch_stable.html # CPU-only build
• https://download.pytorch.org/whl/cu80/torch_stable.html # CUDA 8.0 build
• https://download.pytorch.org/whl/cu90/torch_stable.html # CUDA 9.0 build
• https://download.pytorch.org/whl/cu92/torch_stable.html # CUDA 9.2 build
• https://download.pytorch.org/whl/cu100/torch_stable.html # CUDA 10.0 build

Then, install the file with pip install [downloaded file]

Note: most pytorch versions are available only for specific CUDA versions. For example pytorch=1.0.1 is not available for CUDA 9.2

(Old) PyTorch Linux binaries compiled with CUDA 7.5

These predate the html page above and have to be manually installed by downloading the wheel file and pip install downloaded_file

• cu75/torch-0.3.0.post4-cp36-cp36m-linux_x86_64.whl
• cu75/torch-0.3.0.post4-cp35-cp35m-linux_x86_64.whl
• cu75/torch-0.3.0.post4-cp27-cp27mu-linux_x86_64.whl
• cu75/torch-0.3.0.post4-cp27-cp27m-linux_x86_64.whl
• cu75/torch-0.2.0.post3-cp36-cp36m-manylinux1_x86_64.whl
• cu75/torch-0.2.0.post3-cp35-cp35m-manylinux1_x86_64.whl
• cu75/torch-0.2.0.post3-cp27-cp27mu-manylinux1_x86_64.whl
• cu75/torch-0.2.0.post3-cp27-cp27m-manylinux1_x86_64.whl
• cu75/torch-0.2.0.post2-cp36-cp36m-manylinux1_x86_64.whl
• cu75/torch-0.2.0.post2-cp35-cp35m-manylinux1_x86_64.whl
• cu75/torch-0.2.0.post2-cp27-cp27mu-manylinux1_x86_64.whl
• cu75/torch-0.2.0.post2-cp27-cp27m-manylinux1_x86_64.whl
• cu75/torch-0.2.0.post1-cp36-cp36m-manylinux1_x86_64.whl
• cu75/torch-0.2.0.post1-cp35-cp35m-manylinux1_x86_64.whl
• cu75/torch-0.2.0.post1-cp27-cp27mu-manylinux1_x86_64.whl
• cu75/torch-0.2.0.post1-cp27-cp27m-manylinux1_x86_64.whl
• cu75/torch-0.1.12.post2-cp36-cp36m-linux_x86_64.whl
• cu75/torch-0.1.12.post2-cp35-cp35m-linux_x86_64.whl
• cu75/torch-0.1.12.post2-cp27-none-linux_x86_64.whl
• cu75/torch-0.1.12.post1-cp36-cp36m-linux_x86_64.whl
• cu75/torch-0.1.12.post1-cp35-cp35m-linux_x86_64.whl
• cu75/torch-0.1.12.post1-cp27-none-linux_x86_64.whl
• cu75/torch-0.1.11.post5-cp36-cp36m-linux_x86_64.whl
• cu75/torch-0.1.11.post5-cp35-cp35m-linux_x86_64.whl
• cu75/torch-0.1.11.post5-cp27-none-linux_x86_64.whl
• cu75/torch-0.1.11.post4-cp36-cp36m-linux_x86_64.whl
• cu75/torch-0.1.11.post4-cp35-cp35m-linux_x86_64.whl
• cu75/torch-0.1.11.post4-cp27-none-linux_x86_64.whl
• cu75/torch-0.1.10.post2-cp36-cp36m-linux_x86_64.whl
• cu75/torch-0.1.10.post2-cp35-cp35m-linux_x86_64.whl
• cu75/torch-0.1.10.post2-cp27-none-linux_x86_64.whl
• cu75/torch-0.1.10.post1-cp36-cp36m-linux_x86_64.whl
• cu75/torch-0.1.10.post1-cp35-cp35m-linux_x86_64.whl
• cu75/torch-0.1.10.post1-cp27-none-linux_x86_64.whl
• cu75/torch-0.1.9.post2-cp36-cp36m-linux_x86_64.whl
• cu75/torch-0.1.9.post2-cp35-cp35m-linux_x86_64.whl
• cu75/torch-0.1.9.post2-cp27-none-linux_x86_64.whl
• cu75/torch-0.1.9.post1-cp36-cp36m-linux_x86_64.whl
• cu75/torch-0.1.9.post1-cp35-cp35m-linux_x86_64.whl
• cu75/torch-0.1.9.post1-cp27-none-linux_x86_64.whl
• cu75/torch-0.1.8.post1-cp36-cp36m-linux_x86_64.whl
• cu75/torch-0.1.8.post1-cp35-cp35m-linux_x86_64.whl
• cu75/torch-0.1.8.post1-cp27-none-linux_x86_64.whl
• cu75/torch-0.1.7.post2-cp36-cp36m-linux_x86_64.whl
• cu75/torch-0.1.7.post2-cp35-cp35m-linux_x86_64.whl
• cu75/torch-0.1.7.post2-cp27-none-linux_x86_64.whl
• cu75/torch-0.1.6.post22-cp35-cp35m-linux_x86_64.whl
• cu75/torch-0.1.6.post22-cp27-none-linux_x86_64.whl
• cu75/torch-0.1.6.post20-cp35-cp35m-linux_x86_64.whl
• cu75/torch-0.1.6.post20-cp27-cp27mu-linux_x86_64.whl

Windows binaries

• cpu/torch-1.0.0-cp35-cp35m-win_amd64.whl
• cu80/torch-1.0.0-cp35-cp35m-win_amd64.whl
• cu90/torch-1.0.0-cp35-cp35m-win_amd64.whl
• cu100/torch-1.0.0-cp35-cp35m-win_amd64.whl
• cpu/torch-1.0.0-cp36-cp36m-win_amd64.whl
• cu80/torch-1.0.0-cp36-cp36m-win_amd64.whl
• cu90/torch-1.0.0-cp36-cp36m-win_amd64.whl
• cu100/torch-1.0.0-cp36-cp36m-win_amd64.whl
• cpu/torch-1.0.0-cp37-cp37m-win_amd64.whl
• cu80/torch-1.0.0-cp37-cp37m-win_amd64.whl
• cu90/torch-1.0.0-cp37-cp37m-win_amd64.whl
• cu100/torch-1.0.0-cp37-cp37m-win_amd64.whl
• cpu/torch-0.4.1-cp35-cp35m-win_amd64.whl
• cu80/torch-0.4.1-cp35-cp35m-win_amd64.whl
• cu90/torch-0.4.1-cp35-cp35m-win_amd64.whl
• cu92/torch-0.4.1-cp35-cp35m-win_amd64.whl
• cpu/torch-0.4.1-cp36-cp36m-win_amd64.whl
• cu80/torch-0.4.1-cp36-cp36m-win_amd64.whl
• cu90/torch-0.4.1-cp36-cp36m-win_amd64.whl
• cu92/torch-0.4.1-cp36-cp36m-win_amd64.whl
• cpu/torch-0.4.1-cp37-cp37m-win_amd64.whl
• cu80/torch-0.4.1-cp37-cp37m-win_amd64.whl
• cu90/torch-0.4.1-cp37-cp37m-win_amd64.whl
• cu92/torch-0.4.1-cp37-cp37m-win_amd64.whl

Mac and misc. binaries

For recent macOS binaries, use conda:

e.g.,

conda install pytorch=0.4.1 cuda90 -c pytorch conda install pytorch=0.4.1 cuda92 -c pytorch conda install pytorch=0.4.1 cuda80 -c pytorch conda install pytorch=0.4.1 -c pytorch # No CUDA

• torchvision-0.1.6-py3-none-any.whl
• torchvision-0.1.6-py2-none-any.whl
• torch-1.0.0-cp37-none-macosx_10_7_x86_64.whl
• torch-1.0.0-cp36-none-macosx_10_7_x86_64.whl
• torch-1.0.0-cp35-none-macosx_10_6_x86_64.whl
• torch-1.0.0-cp27-none-macosx_10_6_x86_64.whl
• torch-0.4.0-cp36-cp36m-macosx_10_7_x86_64.whl
• torch-0.4.0-cp35-cp35m-macosx_10_7_x86_64.whl
• torch-0.4.0-cp27-none-macosx_10_7_x86_64.whl
• torch-0.3.1-cp36-cp36m-macosx_10_7_x86_64.whl
• torch-0.3.1-cp35-cp35m-macosx_10_7_x86_64.whl
• torch-0.3.1-cp27-none-macosx_10_7_x86_64.whl
• torch-0.3.0.post4-cp36-cp36m-macosx_10_7_x86_64.whl
• torch-0.3.0.post4-cp35-cp35m-macosx_10_7_x86_64.whl
• torch-0.3.0.post4-cp27-none-macosx_10_7_x86_64.whl
• torch-0.2.0.post3-cp36-cp36m-macosx_10_7_x86_64.whl
• torch-0.2.0.post3-cp35-cp35m-macosx_10_7_x86_64.whl
• torch-0.2.0.post3-cp27-none-macosx_10_7_x86_64.whl
• torch-0.2.0.post2-cp36-cp36m-macosx_10_7_x86_64.whl
• torch-0.2.0.post2-cp35-cp35m-macosx_10_7_x86_64.whl
• torch-0.2.0.post2-cp27-none-macosx_10_7_x86_64.whl
• torch-0.2.0.post1-cp36-cp36m-macosx_10_7_x86_64.whl
• torch-0.2.0.post1-cp35-cp35m-macosx_10_7_x86_64.whl
• torch-0.2.0.post1-cp27-none-macosx_10_7_x86_64.whl
• torch-0.1.12.post2-cp36-cp36m-macosx_10_7_x86_64.whl
• torch-0.1.12.post2-cp35-cp35m-macosx_10_7_x86_64.whl
• torch-0.1.12.post2-cp27-none-macosx_10_7_x86_64.whl
• torch-0.1.12.post1-cp36-cp36m-macosx_10_7_x86_64.whl
• torch-0.1.12.post1-cp35-cp35m-macosx_10_7_x86_64.whl
• torch-0.1.12.post1-cp27-none-macosx_10_7_x86_64.whl
• torch-0.1.11.post5-cp36-cp36m-macosx_10_7_x86_64.whl
• torch-0.1.11.post5-cp35-cp35m-macosx_10_7_x86_64.whl
• torch-0.1.11.post5-cp27-none-macosx_10_7_x86_64.whl
• torch-0.1.11.post4-cp36-cp36m-macosx_10_7_x86_64.whl
• torch-0.1.11.post4-cp35-cp35m-macosx_10_7_x86_64.whl
• torch-0.1.11.post4-cp27-none-macosx_10_7_x86_64.whl
• torch-0.1.10.post1-cp36-cp36m-macosx_10_7_x86_64.whl
• torch-0.1.10.post1-cp35-cp35m-macosx_10_6_x86_64.whl
• torch-0.1.10.post1-cp27-none-macosx_10_7_x86_64.whl
• torch-0.1.9.post2-cp36-cp36m-macosx_10_7_x86_64.whl
• torch-0.1.9.post2-cp35-cp35m-macosx_10_6_x86_64.whl
• torch-0.1.9.post2-cp27-none-macosx_10_7_x86_64.whl
• torch-0.1.9.post1-cp36-cp36m-macosx_10_7_x86_64.whl
• torch-0.1.9.post1-cp35-cp35m-macosx_10_6_x86_64.whl
• torch-0.1.9.post1-cp27-none-macosx_10_7_x86_64.whl
• torch-0.1.8.post1-cp36-cp36m-macosx_10_7_x86_64.whl
• torch-0.1.8.post1-cp35-cp35m-macosx_10_6_x86_64.whl
• torch-0.1.8.post1-cp27-none-macosx_10_7_x86_64.whl
• torch-0.1.7.post2-cp36-cp36m-macosx_10_7_x86_64.whl
• torch-0.1.7.post2-cp35-cp35m-macosx_10_6_x86_64.whl
• torch-0.1.7.post2-cp27-none-macosx_10_7_x86_64.whl
• torch-0.1.6.post22-cp35-cp35m-macosx_10_6_x86_64.whl
• torch-0.1.6.post22-cp27-none-macosx_10_7_x86_64.whl
• torch-0.1.6.post20-cp35-cp35m-linux_x86_64.whl
• torch-0.1.6.post20-cp27-cp27mu-linux_x86_64.whl
• torch-0.1.6.post17-cp35-cp35m-linux_x86_64.whl
• torch-0.1.6.post17-cp27-cp27mu-linux_x86_64.whl
• torch-0.1-cp35-cp35m-macosx_10_6_x86_64.whl
• torch-0.1-cp27-cp27m-macosx_10_6_x86_64.whl
• torch_cuda80-0.1.6.post20-cp35-cp35m-linux_x86_64.whl
• torch_cuda80-0.1.6.post20-cp27-cp27mu-linux_x86_64.whl
• torch_cuda80-0.1.6.post17-cp35-cp35m-linux_x86_64.whl
• torch_cuda80-0.1.6.post17-cp27-cp27mu-linux_x86_64.whl

pytorch最近已經更新到了穩定版本，從Pytorch官網上可以看到有多種安裝方式,比較常見的方法是通過pip和conda安裝，當使用conda安裝的時候，可能會遇到下載過慢的問題，下載的時候可能會遇到無盡的等待。這裏推薦用清華源替代默認conda源的方法，可以解決下載過慢的問題。

添加清華源

    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 --set show_channel_urls yes

另外爲了保險起見，建議同時添加第三方conda源：

    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/cloud/msys2/
    conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/bioconda/
    conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/menpo/
    conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/

根據Python和CUDA選擇對應的版本，然後官方給出提示可通過運行：

conda install pytorch torchvision cudatoolkit=9.0 -c pytorch
但是這裏一定要注意，去掉-c pytorch，安裝的時候纔會默認從清華源下載相應的包，因此這裏用命令行:

conda install pytorch torchvision cudatoolkit=9.0

 

可能遇到的問題

• AttributeError: 'Conv2d' object has no attribute 'padding_mode'

Solution：PyTorch 版本問題，1.1.0 版會出現這個問題，回退到 0.4.0 版本即可。

複製
$ pip install -i https://pypi.tuna.tsinghua.edu.cn/simple --upgrade torch==0.4.0 torchvision==0.2.0