huggingface/transformers

Pull Requests Commits

Merge branch 'main' into add-deepseek-exp

ArthurZucker committed 34 days ago

Verified 296fd445

[Quantization] fix fbgemm (#42561)

MekkCyber committed 34 days ago

Verified 15b79ea8

make sure the FSDP plugin args are appropriately cast to bools (#42566)

winglian committed 34 days ago

Verified 377a8ee7

[Quantization] per tensor quantization kernel (#42560)

MekkCyber committed 34 days ago

Verified 51c5a7a6

[`CI`] Fix copies (#42571)

vasqu committed 34 days ago

Verified a5c061d2

Allow fallback to loading from Auto"SubProcessor".from_pretrained when model_type can't be inferred from config (#42402)

yonigozlan committed 34 days ago

Verified 675e8763

[SAM3] Compute masks once instead of per-layer, fix fa2 crash (#42543)

yonigozlan committed 34 days ago

Verified 80b408d1

Transformers serve fix (#42570)

LysandreJik committed 34 days ago

Verified 52b988d8

Kernel mapping error resolve (#42466)

Aaraviitkgp committed 34 days ago

Verified 3f174109

fix regression (#42569)

SunMarc committed 34 days ago

Verified 53d2bf6d

Fix `parse_response` after tokenizer refactor (#42300)

Rocketknight1 committed 34 days ago

Verified 5690f24e

Add backward compatibility for methods which have been moved to `RotaryEmbeddingConfigMixin` (#42517)

hmellor committed 34 days ago

Verified 0ba8f001

Fix eetq quanto quant methods (#42557)

SunMarc committed 34 days ago

Verified 5efd0d4a

[test] delete `SeamlessM4TProcessorTest::test_save_load_pretrained_addition` (#42562)

ydshieh committed 34 days ago

Verified 4ec83fe9

Fix fp8 + some enhancement (#42455)

SunMarc committed 34 days ago

Verified bc7a268f

[Fix] dots1 expert bias routing (#41663)

fjosw committed 34 days ago

Verified 29e8522b

[Ministral 3] Small fix config (#42537)

patrickvonplaten committed 34 days ago

Verified 64d8cf4f

[Quantization] fix dequant when block size is none & static quantization (#42545)

MekkCyber committed 34 days ago

Verified bb09a30f

Fix failing test in Glm4vMoeIntegrationTest (#42488)

Sai-Suraj-27 committed 34 days ago

Verified ac0769cd

Fix Qwen-VL family with prompt tuning (#42508)

zucchini-nlp committed 35 days ago

Verified 57eeb9ca

Add FastVLM (#41112)

camilla-deckard committed 35 days ago

Verified a6497675

Fix ernie moe (#42535)

vasqu committed 35 days ago

Verified 0fa49db1

[Ministral 3] Add ministral 3 (#42498)

patrickvonplaten committed 35 days ago

Verified bf3f0ae7

[TP plans] Fix some incorrects TP plans (#42448)

Cyrilvallez committed 35 days ago

Verified 6e3f2f8f

small fix tokenizer regex patch (#42528)

ArthurZucker committed 35 days ago

Verified 83fe012d

Fix Qwen3OmniMoE weight init (#42531)

Rocketknight1 committed 35 days ago

Verified dac2ad76

Migration guide typo (#42532)

LysandreJik committed 35 days ago

Verified 063b431d

RC0 Disclaimer (#42530)

LysandreJik committed 35 days ago

Verified 13f8ec70

CLI Migration guide (#42526)

LysandreJik committed 35 days ago

Verified 58f092ec

Fix tied_weights for vlms (#42523)

Cyrilvallez committed 35 days ago

Verified 1b6b6cd0

Older