Distributed Data Paralell stucks after completing one epoch

distributed
1

I have a problem running multigpu training on one node with 2 GPU s.
Currently the program stucks on these lines
imgs = imgs.to(self.device)
true_masks = true_masks.to(self.device)

Can anyone help me?

2

Do you mind providing a script to reproduce your issue, the environment (HW setup, torch version, etc.) that you ran this in, and the logs you are seeing? It would help debugging the issue!