Cómo instalar pytorch en Windows

 Paso 1: instale Python, ejemplo: Python 3.7.9

https://www.python.org/ftp/python/3.7.9/python-3.7.9-amd64.exe

Nota: con Windows 7 debe actualizar:
X64: https://download.microsoft.com/download/F/1/0/F106E158-89A1-41E3-A9B5-32FEB2A99A0B/Windows6.1-KB2533623-x64.msu
X86: https : //download.microsoft.com/download/2/D/7/2D78D0DD-2802-41F5-88D6-DC1D559F206D/Windows6.1-KB2533623-x86.msu

Paso 2: puede crear su propio comando aquí: https://pytorch.org/#pip-install-pytorch

Ejemplo:

1	pip install torch===1.6.0 torchvision===0.7.0 -f https://download.pytorch.org/whl/torch_stable.html

Resultado:

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 dieciséis 17 18 19 20 21 22 23 24 25 26 27 28

C:\Users\Administrator>pip install torch===1.6.0 torchvision===0.7.0 -f https:// download.pytorch.org/whl/torch_stable.html Looking in links: https://download.pytorch.org/whl/torch_stable.html Collecting torch===1.6.0   Downloading https://download.pytorch.org/whl/cu102/torch-1.6.0-cp37-cp37m-win_ amd64.whl (1077.4 MB)      |████████████████████████████████| 1077.4 MB 5.6 kB/s Collecting torchvision===0.7.0   Downloading https://download.pytorch.org/whl/cu102/torchvision-0.7.0-cp37-cp37 m-win_amd64.whl (1.1 MB)      |████████████████████████████████| 1.1 MB 6.8 MB/s Requirement already satisfied: numpy in c:\users\administrator\appdata\local\pro grams\python\python37\lib\site-packages (from torch===1.6.0) (1.15.2) Collecting future   Downloading future-0.18.2.tar.gz (829 kB)      |████████████████████████████████| 829 kB 930 kB/s Requirement already satisfied: pillow>=4.1.1 in c:\users\administrator\appdata\l ocal\programs\python\python37\lib\site-packages (from torchvision===0.7.0) (6.2. 2) Building wheels for collected packages: future   Building wheel for future (setup.py) ... done   Created wheel for future: filename=future-0.18.2-py3-none-any.whl size=491062 sha256=cacf378e9e040cacdaf6012ab0bbad375fa1165b3c74392f6a865eb4c0f4151c   Stored in directory: c:\users\administrator\appdata\local\pip\cache\wheels\56\ b0\fe\4410d17b32f1f0c3cf54cdfb2bc04d7b4b8f4ae377e2229ba0 Successfully built future Installing collected packages: future, torch, torchvision Successfully installed future-0.18.2 torch-1.6.0 torchvision-0.7.0

Nota:
* Descargue el kit de herramientas CUDA 10.2 aquí: https://developer.nvidia.com/cuda-10.2-download-archive?target_os=Windows&target_arch=x86_64

* CUDA 10.2.89 necesita un controlador de gráficos Nvidia> = 440.33

Tabla 1. Kit de herramientas CUDA y versiones de controladores compatibles

Kit de herramientas CUDA	Versión del controlador x86_64
CUDA 11.0 (11.0.171)	> = 450.36.06
CUDA 10.2 (10.2.89)	> = 440,33
CUDA 10.1 (10.1.105)	> = 418,39
CUDA 10.0 (10.0.130)	> = 410,48
CUDA 9.2 (9.2.88)	> = 396,26
CUDA 9.1 (9.1.85)	> = 390,46
CUDA 9.0 (9.0.76)	> = 384,81
CUDA 8.0 (8.0.61 GA2)	> = 375,26
CUDA 8.0 (8.0.44)	> = 367,48
CUDA 7.5 (7.5.16)	> = 352,31
CUDA 7.0 (7.0.28)	> = 346,46