Win10 Python3.6.8 安装 PyTorch

比眉伴天荒 2022-12-29 12:52 136阅读 0赞

**目录**

一、查看CUDA版本

1、cmd命令行查看

2、显卡驱动信息查看

二、安装 PyTorch

1、进入官网

2、选择版本

3、在线安装

4、离线安装

三、其他相关依赖安装

1、PyTorch相关依赖下载

2、相关依赖离线安装

三、验证安装

1、查看已安装依赖

2、查看是否支持GPU

--------------------

## 一、查看CUDA版本 ##

### 1、cmd命令行查看 ###

ncvv -v

nvidia-smi

![watermark_type_ZHJvaWRzYW5zZmFsbGJhY2s_shadow_50_text_Q1NETiBA5pel5pyI5pif6L6wVEVM_size_19_color_FFFFFF_t_70_g_se_x_16][]

### 2、显卡驱动信息查看 ###

**（1.1）win10任务栏搜索NVIDIA**

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxMjYyNTkzNDIx_size_16_color_FFFFFF_t_70][]

**（1.2）点击任务栏左下角“我的电脑”图标，找到 N 开头的软件**

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxMjYyNTkzNDIx_size_16_color_FFFFFF_t_70 1][]

**（1.3）右击任务栏右下角蓝色小图标NVIDIA**

![20201222170553333.png][]

**（2.0）点击 NVIDIA控制面板-帮助-系统信息**

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxMjYyNTkzNDIx_size_16_color_FFFFFF_t_70 2][]

**（3.0）点击 组件，查看 NVCUDA64.DLL对应的驱动版本号**

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxMjYyNTkzNDIx_size_16_color_FFFFFF_t_70 3][]

**（3.1）这里的版本是11.1**

##  ##

## 二、安装 PyTorch ##

PyTorch官网：[PyTorch][]

CUDA官网：[CUDA Toolkit Archive | NVIDIA Developer][CUDA Toolkit Archive _ NVIDIA Developer]

PyTorch版本：[Previous PyTorch Versions | PyTorch][Previous PyTorch Versions _ PyTorch]

PyTorch版本：[https://download.pytorch.org/whl/torch\_stable.html][https_download.pytorch.org_whl_torch_stable.html]

### 1、进入官网 ###

PyTorch官网：[PyTorch][]

### 2、选择版本 ###

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxMjYyNTkzNDIx_size_16_color_FFFFFF_t_70 4][]

### 3、在线安装 ###

pip install torch===1.7.1+cu110 torchvision===0.8.2+cu110 torchaudio===0.7.2 -f https://download.pytorch.org/whl/torch_stable.html

### 4、离线安装 ###

<table>  pytorch离线模块安装  
 <tbody> 
  <tr> 
   <td>功能</td> 
   <td>模块</td> 
   <td>官网</td> 
   <td>安装包</td> 
  </tr> 
  <tr> 
   <td>torch依赖库</td> 
   <td>typing-extensions</td> 
   <td><a href="https://pypi.org/project/typing-extensions/" title="typing-extensions · PyPI" rel="nofollow">typing-extensions · PyPI</a></td> 
   <td><a href="https://files.pythonhosted.org/packages/60/7a/e881b5abb54db0e6e671ab088d079c57ce54e8a01a3ca443f561ccadb37e/typing_extensions-3.7.4.3-py3-none-any.whl" title="typing_extensions-3.7.4.3-py3-none-any.whl&nbsp;" rel="nofollow">typing_extensions-3.7.4.3-py3-none-any.whl&nbsp;</a></td> 
  </tr> 
  <tr> 
   <td>torch依赖库</td> 
   <td>dataclasses</td> 
   <td><a href="https://pypi.org/project/dataclasses/" title="dataclasses · PyPI" rel="nofollow">dataclasses · PyPI</a></td> 
   <td><a href="https://files.pythonhosted.org/packages/fe/ca/75fac5856ab5cfa51bbbcefa250182e50441074fdc3f803f6e76451fab43/dataclasses-0.8-py3-none-any.whl" title="dataclasses-0.8-py3-none-any.whl&nbsp;" rel="nofollow">dataclasses-0.8-py3-none-any.whl&nbsp;</a></td> 
  </tr> 
  <tr> 
   <td>机器学习库</td> 
   <td>torch</td> 
   <td><a href="https://pypi.org/project/torch/" title="torch · PyPI" rel="nofollow">torch · PyPI</a></td> 
   <td><a href="https://files.pythonhosted.org/packages/60/fa/debcf7dcebf446f6ae6ec89177787adefaa1a828027873d961116d9e18f5/torch-1.7.1-cp36-cp36m-win_amd64.whl" title="torch-1.7.1-cp36-cp36m-win_amd64.whl&nbsp;" rel="nofollow">torch-1.7.1-cp36-cp36m-win_amd64.whl&nbsp;</a></td> 
  </tr> 
 </tbody> 
</table>

python -m pip install typing_extensions-3.7.4.3-py3-none-any.whl 
    python -m pip install dataclasses-0.8-py3-none-any.whl 
    python -m pip install torch-1.7.1-cp36-cp36m-win_amd64.whl

## 三、其他相关依赖安装 ##

### 1、PyTorch相关依赖下载 ###

<table>  
 <strong>PyTorch相关依赖</strong>  
 <tbody> 
  <tr> 
   <td>功能</td> 
   <td>模块</td> 
   <td>官网</td> 
   <td>安装包</td> 
  </tr> 
  <tr> 
   <td>jaydebeapi依赖库</td> 
   <td>JPype1</td> 
   <td><a href="https://pypi.org/project/JPype1/" title="JPype1 · PyPI" rel="nofollow">JPype1 · PyPI</a></td> 
   <td><a href="https://files.pythonhosted.org/packages/30/23/064cf1d413877ad597f96b6461a1cb4d4aa47d5edc20157cffd683b1b6d7/JPype1-1.2.0-cp36-cp36m-win_amd64.whl" title="JPype1-1.2.0-cp36-cp36m-win_amd64.whl&nbsp;" rel="nofollow">JPype1-1.2.0-cp36-cp36m-win_amd64.whl&nbsp;</a></td> 
  </tr> 
  <tr> 
   <td>jdbc驱动库</td> 
   <td>jaydebeapi</td> 
   <td><a href="https://pypi.org/project/JayDeBeApi/" title="JayDeBeApi · PyPI" rel="nofollow">JayDeBeApi · PyPI</a></td> 
   <td><a href="https://files.pythonhosted.org/packages/ff/1f/6a627c9bd7dea13235b65fce0fff987507269d41f957c578031796f70319/JayDeBeApi-1.2.3-py3-none-any.whl" title="JayDeBeApi-1.2.3-py3-none-any.whl&nbsp;" rel="nofollow">JayDeBeApi-1.2.3-py3-none-any.whl&nbsp;</a></td> 
  </tr> 
  <tr> 
   <td>matplotlib依赖库</td> 
   <td>six</td> 
   <td><a href="https://pypi.org/project/six" title="six · PyPI" rel="nofollow">six · PyPI</a></td> 
   <td><a href="https://files.pythonhosted.org/packages/ee/ff/48bde5c0f013094d729fe4b0316ba2a24774b3ff1c52d924a8a4cb04078a/six-1.15.0-py2.py3-none-any.whl" title="six-1.15.0-py2.py3-none-any.whl&nbsp;" rel="nofollow">six-1.15.0-py2.py3-none-any.whl&nbsp;</a></td> 
  </tr> 
  <tr> 
   <td>matplotlib依赖库</td> 
   <td>python-dateutil</td> 
   <td><a href="https://pypi.org/project/python-dateutil/" title="python-dateutil · PyPI" rel="nofollow">python-dateutil · PyPI</a></td> 
   <td><a href="https://files.pythonhosted.org/packages/d4/70/d60450c3dd48ef87586924207ae8907090de0b306af2bce5d134d78615cb/python_dateutil-2.8.1-py2.py3-none-any.whl" title="python_dateutil-2.8.1-py2.py3-none-any.whl&nbsp;" rel="nofollow">python_dateutil-2.8.1-py2.py3-none-any.whl&nbsp;</a></td> 
  </tr> 
  <tr> 
   <td>matplotlib依赖库</td> 
   <td>pyparsing</td> 
   <td><a href="https://pypi.org/project/pyparsing" title="pyparsing · PyPI" rel="nofollow">pyparsing · PyPI</a></td> 
   <td><a href="https://files.pythonhosted.org/packages/8a/bb/488841f56197b13700afd5658fc279a2025a39e22449b7cf29864669b15d/pyparsing-2.4.7-py2.py3-none-any.whl" title="pyparsing-2.4.7-py2.py3-none-any.whl&nbsp;" rel="nofollow">pyparsing-2.4.7-py2.py3-none-any.whl&nbsp;</a></td> 
  </tr> 
  <tr> 
   <td>matplotlib依赖库</td> 
   <td>kiwisolver</td> 
   <td><a href="https://pypi.org/project/kiwisolver" title="kiwisolver · PyPI" rel="nofollow">kiwisolver · PyPI</a></td> 
   <td><a href="https://files.pythonhosted.org/packages/6e/df/1250c32ab3b532c32a7e47c1cd240faba98f75b1b5150939b10e9bffb758/kiwisolver-1.3.1-cp36-cp36m-win_amd64.whl" title="kiwisolver-1.3.1-cp36-cp36m-win_amd64.whl&nbsp;" rel="nofollow">kiwisolver-1.3.1-cp36-cp36m-win_amd64.whl&nbsp;</a></td> 
  </tr> 
  <tr> 
   <td>matplotlib依赖库</td> 
   <td>cycler</td> 
   <td><a href="https://pypi.org/project/cycler" title="Cycler · PyPI" rel="nofollow">Cycler · PyPI</a></td> 
   <td><a href="https://files.pythonhosted.org/packages/f7/d2/e07d3ebb2bd7af696440ce7e754c59dd546ffe1bbe732c8ab68b9c834e61/cycler-0.10.0-py2.py3-none-any.whl" title="cycler-0.10.0-py2.py3-none-any.whl&nbsp;" rel="nofollow">cycler-0.10.0-py2.py3-none-any.whl&nbsp;</a></td> 
  </tr> 
  <tr> 
   <td>2D绘图库</td> 
   <td>matplotlib&nbsp;&nbsp; &nbsp;</td> 
   <td><a href="https://pypi.org/project/matplotlib/" title="matplotlib · PyPI" rel="nofollow">matplotlib · PyPI</a></td> 
   <td><a href="https://files.pythonhosted.org/packages/39/76/9959c81f3b0774db4052fcf2121c8220bdc1d111212dd2692d90fc1c6471/matplotlib-3.3.3-cp36-cp36m-win_amd64.whl" title="matplotlib-3.3.3-cp36-cp36m-win_amd64.whl&nbsp;" rel="nofollow">matplotlib-3.3.3-cp36-cp36m-win_amd64.whl&nbsp;</a></td> 
  </tr> 
  <tr> 
   <td>mysql客户端库</td> 
   <td>pymysql</td> 
   <td><a href="https://pypi.org/project/PyMySQL/" title="PyMySQL · PyPI" rel="nofollow">PyMySQL · PyPI</a></td> 
   <td> <p><a href="https://files.pythonhosted.org/packages/1a/ea/dd9c81e2d85efd03cfbf808736dd055bd9ea1a78aea9968888b1055c3263/PyMySQL-0.10.1-py2.py3-none-any.whl" title="PyMySQL-0.10.1-py2.py3-none-any.whl" rel="nofollow">PyMySQL-0.10.1-py2.py3-none-any.whl</a></p> </td> 
  </tr> 
 </tbody> 
</table>

### 2、相关依赖离线安装 ###

python -m pip install JPype1-1.2.0-cp36-cp36m-win_amd64.whl 
    python -m pip install JayDeBeApi-1.2.3-py3-none-any.whl 
    python -m pip install six-1.15.0-py2.py3-none-any.whl 
    python -m pip install python_dateutil-2.8.1-py2.py3-none-any.whl 
    python -m pip install pyparsing-2.4.7-py2.py3-none-any.whl 
    python -m pip install kiwisolver-1.3.1-cp36-cp36m-win_amd64.whl 
    python -m pip install cycler-0.10.0-py2.py3-none-any.whl 
    python -m pip install matplotlib-3.3.3-cp36-cp36m-win_amd64.whl 
    python -m pip install PyMySQL-0.10.1-py2.py3-none-any.whl

## 三、验证安装 ##

### 1、查看已安装依赖 ###

pip3 list | findstr torch

![watermark_type_ZHJvaWRzYW5zZmFsbGJhY2s_shadow_50_text_Q1NETiBA5pel5pyI5pif6L6wVEVM_size_12_color_FFFFFF_t_70_g_se_x_16][]

### 2、查看是否支持GPU ###

import torch
    
    print(torch.cuda)
    print(torch.__version__)
    print(torch.cuda.is_available())

![watermark_type_ZHJvaWRzYW5zZmFsbGJhY2s_shadow_50_text_Q1NETiBA5pel5pyI5pif6L6wVEVM_size_20_color_FFFFFF_t_70_g_se_x_16][]

[watermark_type_ZHJvaWRzYW5zZmFsbGJhY2s_shadow_50_text_Q1NETiBA5pel5pyI5pif6L6wVEVM_size_19_color_FFFFFF_t_70_g_se_x_16]: /images/20221120/0c3b36835ae8415c9a64cd141f0bd09b.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxMjYyNTkzNDIx_size_16_color_FFFFFF_t_70]: /images/20221120/08ab700417314bbda1b5869eaa7f4669.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxMjYyNTkzNDIx_size_16_color_FFFFFF_t_70 1]: https://img-blog.csdnimg.cn/20201222170411422.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxMjYyNTkzNDIx,size_16,color_FFFFFF,t_70
[20201222170553333.png]: https://img-blog.csdnimg.cn/20201222170553333.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxMjYyNTkzNDIx_size_16_color_FFFFFF_t_70 2]: https://img-blog.csdnimg.cn/20201222170520404.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxMjYyNTkzNDIx,size_16,color_FFFFFF,t_70
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxMjYyNTkzNDIx_size_16_color_FFFFFF_t_70 3]: https://img-blog.csdnimg.cn/20201222170454969.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxMjYyNTkzNDIx,size_16,color_FFFFFF,t_70
[PyTorch]: https://pytorch.org/
[CUDA Toolkit Archive _ NVIDIA Developer]: https://developer.nvidia.com/cuda-toolkit-archive
[Previous PyTorch Versions _ PyTorch]: https://pytorch.org/get-started/previous-versions/
[https_download.pytorch.org_whl_torch_stable.html]: https://download.pytorch.org/whl/torch_stable.html
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxMjYyNTkzNDIx_size_16_color_FFFFFF_t_70 4]: https://img-blog.csdnimg.cn/2020122217072879.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxMjYyNTkzNDIx,size_16,color_FFFFFF,t_70
[watermark_type_ZHJvaWRzYW5zZmFsbGJhY2s_shadow_50_text_Q1NETiBA5pel5pyI5pif6L6wVEVM_size_12_color_FFFFFF_t_70_g_se_x_16]: https://img-blog.csdnimg.cn/20210910111047914.png?x-oss-process=image/watermark,type_ZHJvaWRzYW5zZmFsbGJhY2s,shadow_50,text_Q1NETiBA5pel5pyI5pif6L6wVEVM,size_12,color_FFFFFF,t_70,g_se,x_16
[watermark_type_ZHJvaWRzYW5zZmFsbGJhY2s_shadow_50_text_Q1NETiBA5pel5pyI5pif6L6wVEVM_size_20_color_FFFFFF_t_70_g_se_x_16]: https://img-blog.csdnimg.cn/20210910112142343.png?x-oss-process=image/watermark,type_ZHJvaWRzYW5zZmFsbGJhY2s,shadow_50,text_Q1NETiBA5pel5pyI5pif6L6wVEVM,size_20,color_FFFFFF,t_70,g_se,x_16