[deepspeed] fix resume from checkpoint (#11352)

This PR fixes a bug that most likely somehow got exposed (not caused) by https://github.com/huggingface/transformers/pull/11318 - surprisingly the same test worked just fine before that other PR.
This commit is contained in:
Stas Bekman 2021-04-21 07:48:15 -07:00 committed by GitHub
parent 74712e22f3
commit ca7ff64f5b
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23

View File

@ -1017,7 +1017,7 @@ class Trainer:
"yield to errors or unwanted behaviors."
)
if self.deepspeed:
if args.deepspeed:
# will be resumed in deepspeed_init
pass
else: