transformers
Add the auto_find_batch_size capability from Accelerate into Trainer
#17068
Merged
Merged
Add the auto_find_batch_size capability from Accelerate into Trainer #17068
muellerzr
merged
32 commits into
main
from
muellerzr-memory-decorator
Start implementation
297e7d11
Fix partial
3d783720
change where logic happens
c78ed025
Change how decorator is called
a933e182
Should fix issue
0ad8a778
Style
c5d5d7b9
Proper style
8da6c7e5
Add tests
f7595359
Add to deps and remove guard
f97111a5
Try now?
253f0cbf
Dep tables
4f169862
Rm import
5624504c
Add test
6dcb9df8
Update setup.py to use latest
48753df6
Fixup dep style
ab12e995
Move auto_find_batch_size to a TrainingArgument
81fe4d60
Use requires_backends, add decorators to tests
3a777fbe
move import
56b884d8
Rework test to use glue
b12ab6a8
Fix import errors
4ce0683a
Working trainer test, still need to fix utils
77528130
Add accelerate to depslist
f8e89278
Update config with install accelerate
88d20c19
sgugger
commented
on 2022-05-04
Restructure inner loop
bdc46922
Slow
97b575ea
sgugger
approved these changes
on 2022-05-04
Restyle
ff6caca9
Fix logging import
62d9a22b
Add stas fix to stages
5c6c4b22
Style
ef92961d
Finally clean
475e164b
LysandreJik
approved these changes
on 2022-05-09
Merge branch 'main' into muellerzr-memory-decorator
00d316b0
Clean
ada7bd1e
Login to write a write a comment.
Login via GitHub