How to immediately end training?

[vedantroy]

I know you can end training by returning -1 in on_train_batch_start, or whatever the callback is called, but this can take a while to finish if you have a lot of epochs.

Is there a way to tell Pytorch Lightning to immediately end training?

[treblenalto]

If your purpose is to debug training process, try Trainer(fast_dev_run=True) or you can shorten epochs by using fraction of data with Trainer(limit_train_batches=0.1, limit_val_batches=0.01)