huggingface/transformers

Pull Requests Commits

Release: v4.51.3

LysandreJik committed 272 days ago

5f4ecf2d

Handle torch ver in flexattn (#37400)

Kh4L committed 272 days ago

befb5344

handle torch version edge cases (#37399)

winglian committed 272 days ago

bffeefe4

Add glm4 (#37388)

ArthurZucker committed 272 days ago

5c076fb4

Attention Quantization with FBGemm & TP (#37384)

MekkCyber committed 276 days ago

28c9541c

use `rms_norm_eps` for the L2Norm for Llama4 (#37418)

ArthurZucker committed 276 days ago

56512077

mark llama4 as not supported with fa2 (#37416)

winglian committed 276 days ago

ac9c7f37

Fix Llama4 offset (#37414)

Cyrilvallez committed 276 days ago

894783c9

ArthurZucker committed 277 days ago

eae06ff6

the fix that did not get in (#37370)

ArthurZucker committed 277 days ago

3148a41f

Multiple llama4 fixe (#37353)

ArthurZucker committed 278 days ago

10baffb5

ArthurZucker committed 278 days ago

4a88ffae

Fixing flex attention for torch=2.6.0 (#37285)

salmanmohammadi committed 278 days ago

f19aec73

more fixes for post-training llama4 (#37329)

winglian committed 278 days ago

d8f0695e

Remove HQQ from caching allocator warmup (#37347)

Cyrilvallez committed 278 days ago

d27c8c38

fix derived berts `_init_weights` (#37341)

Cyrilvallez committed 278 days ago

04c0cedc

Fix `init empty weights` without accelerate (#37337)

Cyrilvallez committed 278 days ago

4f536ba0

Fix deepspeed with quantization (#37324)

Cyrilvallez committed 278 days ago

6b82af0a

fix llama4 training (#37319)

hiyouga committed 278 days ago

2bf3d4ac

fix flex attn when optional args aren't passed (#37327)

winglian committed 278 days ago

a79b7abe

Release: v4.51.0

LysandreJik committed 281 days ago

0720e206

Add llama4 (#37307)

ArthurZucker committed 281 days ago

Verified 25b7f272

Hf Xet extra (#37305)

LysandreJik committed 281 days ago

Verified aa40fda3

Fix deepspeed loading (part 2) (#37306)

Cyrilvallez committed 281 days ago

Verified e9457158

Fix deepspeed loading (#37281)

Cyrilvallez committed 281 days ago

Verified 84aa13dd

Update OpenAI GPT model card (#37255)

linnettuscano committed 281 days ago

Verified 0ef339ff

Updated T5 model card with standardized format (#37261)

ShararehY committed 281 days ago

Verified 46d73910

Updated model card for distilbert (#37157)

ChathuminaVimukthi committed 281 days ago

Verified 579135a2

mobilebert model card update (#37256)

Reshan123 committed 281 days ago

Verified 8cd57eb7

Fix: Unexpected Keys, Improve `run_compressed`, Rename Test Folder (#37077)

rahul-tuli committed 282 days ago

Verified ebe47ce3

Older