transformers
fixes to properly shard FSDP across cpu and meta for cpu_efficient_loading for prequantized 4bit
#32276
Merged
Overview
Commits
3
Changes
View On GitHub

fixes to properly shard FSDP across cpu and meta for cpu_efficient_loading for prequantized 4bit #32276

SunMarc merged 3 commits into huggingface:main from winglian:fsdp-meta-sharding
winglian
winglian fixes to properly shard FSDP across cpu and meta for cpu_efficient_lo…
0fc478b9
matthewdouglas
matthewdouglas
matthewdouglas commented on 2024-07-28
HuggingFaceDocBuilderDev
muellerzr
muellerzr approved these changes on 2024-07-29
muellerzr muellerzr requested a review from LysandreJik LysandreJik 1 year ago
winglian chore: lint
aec71876
winglian
winglian commented on 2024-07-29
matthewdouglas matthewdouglas added PyTorch FSDP
winglian use existing is_local_dist_rank_0 fn to check devices
64751e54
LysandreJik
LysandreJik approved these changes on 2024-07-30
SunMarc
SunMarc approved these changes on 2024-07-30
SunMarc SunMarc merged 62c60a30 into main 1 year ago
winglian winglian deleted the fsdp-meta-sharding branch 1 year ago

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
PyTorch FSDP
Milestone
No milestone
Terms Of Use Privacy Policy Refund Policy Impressum