num_ Works parameters   This is the usage of the process.

 
num_ When works is nonzero, the for loop will be stuck here.

 
Changed to 0, ready to run.

Add the num parameter in the dataloader_ Workers can be set to 0

Resources
debugger freezes stepping forward when using torch with workers (multiprocessing)
Zhihu – torch dataloader uses batch and num_ What is the principle of works parameter?Principle analysis
CSDN – pytorch training encountered in the stuck stop and other problems (recommended)
using the python dataloader stuck error solution
dataloader, when num_ worker > 0, there is a bug
to solve the pytorch dataloader num_ Problems of workers