Training on the multi-GPUs but stuck in loss.backward()

[winchest (ZHAO)]

OMGGG！Finally I find that the os.environ code part seems to be conflict with torch.nn.DataParallel.So when I delete the os.environ code It worked!!