transformers
FSDP grad accum fix
#34645
Merged

FSDP grad accum fix #34645

winglian
winglian add gradient accumulation steps tests for fsdp
95b718b1
winglian invert no_sync context to fix training for fsdp
edd102fc
winglian
muellerzr
muellerzr approved these changes on 2024-11-07
muellerzr muellerzr requested a review from ArthurZucker ArthurZucker 1 year ago
winglian
SunMarc
SunMarc approved these changes on 2024-11-15
SunMarc SunMarc requested a review from ydshieh ydshieh 1 year ago
HuggingFaceDocBuilderDev
ydshieh
ydshieh approved these changes on 2024-11-15
ArthurZucker
ArthurZucker approved these changes on 2024-11-15
ArthurZucker ArthurZucker merged b0c0ba7b into main 1 year ago

Login to write a write a comment.

Login via GitHub

Assignees
No one assigned
Labels
Milestone