使用 pytorch 进行训练时，调试器挂起，即使运行正常

Question

Trying to train with pytorch hangs in debug mode, but works in run mode.

    sampler_train = WeightedRandomSampler(
        sample_weights_train,
        num_samples=len(sample_weights_train),
        replacement=True
    )

    train_loader = torch.utils.data.DataLoader(
        train_set,
        sampler=sampler_train,
        batch_size=32,
        num_workers=2
    )


   for epoch in range(10): 
        for i, data in enumerate(train_loader, 0):
            model.train()
            print("something")

After placing a breakpoint on model.train() , then moving on to the next line, "something" is never printed in debug mode, but is printed in run mode in Pycharm.

How to debug my code?

Answer 1

After a long search, I found the answer here , which led to here

Setting Gevent Compatible in Preferences | Build, Execution, Deployment | Python Debugger solves the issue.