Installing pytorch under Anaconda

---

Preface

python It really gives me a headache

---

One 、anaconda Source

anaconda The source of may have been a wall , Use anaconda The installation environment should hit the wall . Those who understand are welcome to show the way

Will be displayed :CONDA HTTP ERROR
There are also tutorials for configuring Tsinghua source on the Internet , But I can't find anaconda Of anacord The configuration file .

Two 、 download pytorch Of whl file

1、 Download url ：

pytorch Official website

2. Selection Guide

1. build Just like yourself
2. The operating system should be based on your own system
   If a child doesn't know his or her own operating system
   Press down win key , Open Settings , Here's the picture
   
   I am a win64
3. package choice pip, Easy to install , Ask is that I am skilled
4. Language choice python, I just want to python Fuck her
5. platform Look at the platform , As for how to look at , With windows For example , To ask is to use only windows
   Open the control panel , Search for nvidia
   
   open , Choose help - system information - Components
   
   The third line is cuda edition , I am a cuda11.0
   If you happen to have the version above the picture
   

Just copy run the command This line of text .
without , that

3. Choose the right version

Click on PREVIOUS VERSION of Pytorch
Then find what you want in it pytorch The version of the wheel file

Find your own version , Then copy the text

3、 ... and . install

open anaconda prompt, Enter the copied text and press enter

Four . follow-up

Wrong report , Tips MEMORY ERROR, It should be a memory error
because pip The installed caching mechanism wants to read the entire file into memory before starting the installation , This may result in insufficient memory .
Modify the command

pip install --no-cache-dir torch==1.x.x ...