Merge pull request #882 from ShoufaChen/master

fix `use_amp`

Author: rwightman

train.py

@@ -397,7 +397,7 @@ def main():
     # setup synchronized BatchNorm for distributed training
     if args.distributed and args.sync_bn:
         assert not args.split_bn
-        if has_apex and use_amp != 'native':
+        if has_apex and use_amp == 'apex':
             # Apex SyncBN preferred unless native amp is activated
             model = convert_syncbn_model(model)
         else:
@@ -451,7 +451,7 @@ def main():
 
     # setup distributed training
     if args.distributed:
-        if has_apex and use_amp != 'native':
+        if has_apex and use_amp == 'apex':
             # Apex DDP preferred unless native amp is activated
             if args.local_rank == 0:
                 _logger.info("Using NVIDIA APEX DistributedDataParallel.")