PR #26180 FSDP tests and checkpointing fixes

FSDP tests and checkpointing fixes #26180

pacman100 merged 26 commits into main from smangrul/fsdp-tests

add fsdp tests

1045c925

Update test_fsdp.py

f172658f

Update test_fsdp.py

498f4a15

fixes

141dce52

checks

98c608df

Update trainer.py

986c8b5b

fix

4239fa11

fixes for saving/resuming checkpoints

b61509e3

fixes

085bcb37

add tests and delete debug statements

85357d35

fixing tests

816ccb2b

Update test_fsdp.py

c3e0a1a6

fix tests

8d57ed82

fix tests

c4f8b685

minor nits

c0fc9300

fix code style and quality

dd2431b1

refactor and modularize test code

6cb7db30

reduce the time of tests

5ae3309c

reduce the test time

28a0300c

fix test

7ec28e7e

reduce test time

56dfb476

reduce test time

28cd21d1

pacman100 requested a review from

muellerzr 2 years ago

pacman100 requested a review from

ArthurZucker 2 years ago

pacman100 marked this pull request as ready for review 2 years ago

fix failing tests

a6c27b2d

fix

94a5e0a5

pacman100 requested a review from

BenjaminBossan 2 years ago

ArthurZucker approved these changes on 2023-09-15

Apply suggestions from code review

a80385a9

resolve comments

eb7cb98a

pacman100 requested a review from

ArthurZucker 2 years ago

ArthurZucker approved these changes on 2023-09-19

pacman100 merged 382ba670 into main 2 years ago

pacman100 deleted the smangrul/fsdp-tests branch 2 years ago

Reviewers

ArthurZucker

muellerzr

BenjaminBossan

Assignees

No one assigned

Labels

None yet

Milestone

No milestone

transformers FSDP tests and checkpointing fixes #26180 Merged

FSDP tests and checkpointing fixes #26180

transformers
FSDP tests and checkpointing fixes
#26180

Merged