transformers
Much more efficient and clear weight initialization and tie weights
#42191
Merged

Much more efficient and clear weight initialization and tie weights #42191

Cyrilvallez merged 45 commits into main from better-init-2
Cyrilvallez
HuggingFaceDocBuilderDev
ArthurZucker
ArthurZucker commented on 2025-11-14
Cyrilvallez Cyrilvallez changed the title Much more efficient and clear weight initialization Much more efficient and clear weight initialization and tie weights 69 days ago
Cyrilvallez everything untilo informer
da26896e
Cyrilvallez everything until perceiver
d561c6f9
Cyrilvallez all of them finally
ceea3058
Cyrilvallez style
187bb8ef
Cyrilvallez replace by transformers init everywhere
2cd2addc
Cyrilvallez use relative import instead
6bdffed5
Cyrilvallez deprecated models
d25fe728
Cyrilvallez style
82899acb
Cyrilvallez start contexts
a4ab5985
Cyrilvallez small fixes
192151e0
Cyrilvallez fix modular
5efa9a8b
Cyrilvallez remove class switch
c882d608
Cyrilvallez do not initialize tied weights
22a55a36
Cyrilvallez typo
694440bb
Cyrilvallez fix
5a0174ec
Cyrilvallez improve
5423e064
Cyrilvallez improve comments
9b7ace53
Cyrilvallez improve
4acef54a
Cyrilvallez improve
c58d243c
Cyrilvallez fix zamba
2edc8c17
Cyrilvallez fix import
2f40139d
Cyrilvallez add the post_init
2dd4e00a
Cyrilvallez more post_init
3ede2872
Cyrilvallez Cyrilvallez force pushed from f33c91ec to 3ede2872 69 days ago
Cyrilvallez fix
86f7169d
Cyrilvallez protect
706799e9
ArthurZucker
ArthurZucker commented on 2025-11-14
Cyrilvallez more post_init
1da2d273
Cyrilvallez fix
83e0ada2
Cyrilvallez fixes
50187a90
Cyrilvallez fix
16173f06
Cyrilvallez fix
bae372ae
Cyrilvallez switch flag name
8500bcf9
Cyrilvallez more fixes
cdada869
Cyrilvallez fixes
99961fc6
Cyrilvallez fixes
557ef759
Cyrilvallez Cyrilvallez force pushed from 79e84f90 to 557ef759 69 days ago
Cyrilvallez Merge branch 'main' into better-init-2
2dd08170
Cyrilvallez copies
912440bc
Cyrilvallez fix
acdaf9e9
Cyrilvallez finally find the culprit
cc10ea4e
Cyrilvallez style
627e77b3
Cyrilvallez last small
db42923c
ArthurZucker
ArthurZucker approved these changes on 2025-11-14
Cyrilvallez big bird
17115a22
Cyrilvallez better
bbdc5a5b
Cyrilvallez update init check
3a12aec8
github-actions
Cyrilvallez final touch
9beb88c0
Cyrilvallez do it everywhere
60928045
Cyrilvallez Cyrilvallez merged 8598421b into main 68 days ago
Cyrilvallez Cyrilvallez deleted the better-init-2 branch 68 days ago

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
Milestone