huggingface/transformers

Pull Requests Commits

use torch primitives

Cyrilvallez committed 302 days ago

Verified f3e202e0

Fix/best model checkpoint fix (#35885)

seanswyi committed 302 days ago

Verified 691d1b52

[model loading] don't `gc.collect()` if only 1 shard is used (#36721)

gante committed 302 days ago

Verified 3bd1a0dd

Cleanup the regex used for doc preprocessing (#36648)

Rocketknight1 committed 302 days ago

Verified 8cb522b4

Make the flaky list a little more general (#36704)

Rocketknight1 committed 302 days ago

Verified 72861e11

Gemma3 processor typo (#36710)

Kuangdd01 committed 302 days ago

Verified 53742b11

Add support for fast image processors in add-new-model-like CLI (#36313)

yonigozlan committed 303 days ago

Verified 69bc8484

Final CI cleanup (#36703)

Rocketknight1 committed 303 days ago

Verified 48ef468c

Add GGUF support to T5-Encoder (#36700)

Isotr0py committed 303 days ago

Verified b070025a

Handling an exception related to HQQ quantization in modeling (#36702)

MekkCyber committed 303 days ago

Verified 4a60bae8

fix: fsdp sharded state dict wont work for save_only_model knob (#36627)

kmehant committed 303 days ago

Verified 09a309d2

Add loading speed test (#36671)

Cyrilvallez committed 303 days ago

Verified 2a004f9f

[CI] Automatic rerun of certain test failures (#36694)

gante committed 303 days ago

Verified a3201cea

chore: fix typos in utils module (#36668)

threewebcode committed 303 days ago

Verified d8456938

Fix dtype for params without tp_plan (#36681)

Cyrilvallez committed 303 days ago

Verified 32c95bd8

fix type annotation for ALL_ATTENTION_FUNCTIONS (#36690)

WineChord committed 303 days ago

Verified bb965d8e

Change Qwen2_VL image processors to have init and call accept the same kwargs (#36207)

yonigozlan committed 303 days ago

Verified 1c287aec

Upgrading torch version and cuda version in quantization docker (#36264)

MekkCyber committed 303 days ago

Verified 65b8e38a

fix wandb hp search unable to resume from sweep_id (#35883)

bd793fcb committed 303 days ago

Verified 87b30c35

Changing the test model in Quanto kv cache (#36670)

MekkCyber committed 303 days ago

Verified 47cc4da3

Fix slicing for 0-dim param (#36580)

SunMarc committed 303 days ago

Verified bc3d5781

Update config.torch_dtype correctly (#36679)

SunMarc committed 303 days ago

Verified fbb18ce6

[Cache] Don't initialize the cache on `meta` device (#36543)

gante committed 303 days ago

Verified c4161238

Fix rescale normalize inconsistencies in fast image processors (#36388)

yonigozlan committed 303 days ago

Verified 79254c9b

Refactor siglip2 fast image processor (#36406)

yonigozlan committed 303 days ago

Verified 48292a98

Remove differences between init and preprocess kwargs for fast image processors (#36186)

yonigozlan committed 303 days ago

Verified ea219ed1

[quants] refactor logic for modules_to_not_convert (#36672)

SunMarc committed 303 days ago

Verified cc3a361b

Remove hardcoded slow image processor class in processors supporting fast ones (#36266)

yonigozlan committed 303 days ago

Verified bc3253f0

Fix Failing GPTQ tests (#36666)

MekkCyber committed 304 days ago

Verified 0013ba61

Don't accidentally mutate the base_model_tp_plan (#36677)

Rocketknight1 committed 304 days ago

Verified c7eb9558

Older