Deploy the project to GPU and run

1. Set up my running environment

2. Log in to my GPU

3. In my GPU Select the corresponding environment

Choose the virtual environment you want to run python

Select the folder you want to synchronize , Two boxes will appear , The front box is your local directory , The latter is the directory you want to upload .

4. see GPU Every piece on the GPU The capacity of

Enter the code

 nvidia-smi

If your video memory is small , There may be a mistake

RuntimeError:CURD out of memory

At this time, you need to choose the larger part of the video memory GPU function , Or turn down your own batc_size

#  Appoint GPU Run my project 
import os
os.environ["CUDA_VISIBLE_DEVICES"] = "0"

The last number 0 Indicates using the 0 block GPU

5. Synchronous data

Method of synchronizing data :

6. When you run someone else's code, you report an error path error