maxmustermann/DeepLearningExamples
1
0
Fork 0
Code Issues Pull requests Projects Releases Wiki Activity

Fix a logic flaw that causes checkpoints to be saved at wrong epochs.

Browse source
This commit is contained in:
Peter-1213 2021-10-27 01:22:33 +08:00
parent 9b86d8c0b8
commit 9cf6c3824b
1 changed files with 1 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
PyTorch/Translation/Transformer/train.py
View file

@ -364,7 +364,7 @@ def _all_gather_predictions(predictions):
def save_checkpoint(args, trainer, epoch_itr, val_loss):
if epoch_itr.epoch % args.save_interval == 0:
if epoch_itr.epoch % args.save_interval != 0:
return
if args.no_save or not distributed_utils.is_master(args):
return