DeepSpeed
Switch modeling to use transformers and torch version for BERT
#4329
Open
Open
Switch modeling to use transformers and torch version for BERT #4329
loadams
wants to merge
7 commits into
master
from
loadams/remove-modeling
Remove modeling and modeling.py files in facor of transformers equiva…
9cea1d1c
Replace with versions from transformers and torch
fa9fe945
Update to match the parameters used by Bertencoder
3ad18557
Merge branch 'master' into loadams/remove-modeling
9f04e777
Update to correct signautre
f9ad1a49
Merge branch 'master' into loadams/remove-modeling
8159a2df
Make backward look like forward?
46f5e725
Login to write a write a comment.
Login via GitHub