transformers
Update form pretrained to make TP a first class citizen
#36335
Merged

Update form pretrained to make TP a first class citizen #36335

ArthurZucker merged 79 commits into main from safe-tensors
ArthurZucker
ArthurZucker clean code
4fdb6a47
ArthurZucker Merge branch 'main' of github.com:huggingface/transformers into safe-…
fe79aef9
ArthurZucker oups
7f0ca674
ArthurZucker fix merge
db4f78bc
ArthurZucker yups
95753329
ArthurZucker fix if
7346376b
ArthurZucker now you can play
3fe75a0a
ArthurZucker fix shape issue
034beab0
ArthurZucker try non blocking
7be41570
ArthurZucker fix
82a471fc
HuggingFaceDocBuilderDev
ArthurZucker updates
fbf2912d
ArthurZucker up
60824df6
ArthurZucker updates
4981ffc3
ArthurZucker fix most of thetests
1be42fc8
ArthurZucker update
995f2250
ArthurZucker update
85714010
ArthurZucker small updates
b1ee64c4
ArthurZucker up
a78842a7
ArthurZucker fix the remaining bug?
d92b4e34
ArthurZucker update
d573acbc
ArthurZucker Merge branch 'main' into safe-tensors
740b52bb
ArthurZucker rename when you read from the file
25bb5694
ArthurZucker Merge branch 'safe-tensors' of github.com:huggingface/transformers in…
efb11165
ArthurZucker buffer issues
d8aa45fc
ArthurZucker current status
6a352480
ArthurZucker cleanup
8e3a6aeb
ArthurZucker properly allocate dumb memory
dfc864f9
ArthurZucker update a small bug
a08c849b
ArthurZucker fix colwise rep issue
2c7ab616
ArthurZucker fix keep in float 32 that was keeping everything in float 32
7efe2192
ArthurZucker typo
179e26b5
ArthurZucker more fixes with keep_in_fp32_modules as we use to serach on it
046d6a11
ArthurZucker ArthurZucker changed the title Update form pretrained Update form pretrained to make TP a first class citizen 1 year ago
ArthurZucker fix ROPE dtype for TP
52eda207
ArthurZucker remove what's breaking the tests
ae79fad4
ArthurZucker updates
acb45d69
ArthurZucker update and fixes
93555c05
ArthurZucker ArthurZucker marked this pull request as ready for review 1 year ago
ArthurZucker Merge branch 'main' into safe-tensors
c14ecccc
ArthurZucker Merge branch 'main' of github.com:huggingface/transformers into safe-…
71fe6725
ArthurZucker Merge branch 'safe-tensors' of github.com:huggingface/transformers in…
c3c6d85a
ArthurZucker small cleanup after merging
f140cc82
ArthurZucker allocate 2x to be safe
1fdd5225
ArthurZucker style, auto
11b11077
ArthurZucker update
d5c60231
ArthurZucker yup nit
0b3a18b7
ArthurZucker fix
d224cf88
ArthurZucker remove slow as fuck torch api :(
f6893c90
SunMarc
SunMarc commented on 2025-02-25
ArthurZucker work
18988233
ArthurZucker fixup
4c2087f5
ArthurZucker update
3e2526ef
ArthurZucker brting the fix back
42c6119e
ArthurZucker fix and update
45225051
ArthurZucker fixes
6b9f2435
ArthurZucker updates because some suggestions were wrong :eyes:
752bc950
ArthurZucker update?
a5b84ec3
ArthurZucker fuck this bloated function
b53d381c
ArthurZucker typo
0c4e1731
ArthurZucker fix the dumb prefix thing once and forall
4e8a8d57
ArthurZucker fixes here and there
a9adbeb5
ArthurZucker updates
feaf7f14
ArthurZucker remove prints
a0b7af41
ArthurZucker Merge branch 'main' of github.com:huggingface/transformers into safe-…
4e03eb9e
ArthurZucker fix strict cases
0af4c229
ArthurZucker styel
366aa1ff
ArthurZucker properly fix keys on load!
634016ae
ArthurZucker update
95fd0018
ArthurZucker fix base model prefix issue
640dc389
ArthurZucker style
e6bbf628
ArthurZucker update
13522e9c
ArthurZucker fix all?
d85e53f1
ArthurZucker remoce 1 print
750c04cb
ArthurZucker fix the final etsts
31b90df5
ArthurZucker fixup
796cfb73
ArthurZucker last nits
0e2eca8e
ArthurZucker fix the detach issue which cause a 2x slowdown
7cab57ed
ArthurZucker fixup
3c3a51e4
ArthurZucker small fixes
050425c0
ArthurZucker ultra nit
28245306
ArthurZucker fix
62ed1be6
ArthurZucker fix
c049e790
ArthurZucker ArthurZucker added Tensor Parallel
ArthurZucker ArthurZucker merged 1603018e into main 1 year ago
ArthurZucker ArthurZucker deleted the safe-tensors branch 1 year ago

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
Milestone