huggingface/transformers

Pull Requests Commits

Add AutoRound quantization support (#37393)

wenhuach21 committed 254 days ago

Verified b3492ff9

Correct warm-up with fp8 (#37670)

Cyrilvallez committed 254 days ago

Verified 96089086

Fix duplicated weights in fp8 quantization (#37667)

Cyrilvallez committed 254 days ago

Verified 6614209b

[qwen-omni] fix training (#37517)

zucchini-nlp committed 255 days ago

Verified dcf6df5b

Introduce GradientCheckpointingLayer (#37223)

qubvel committed 255 days ago

Verified 9167fada

Fixes #37219 : RecurrentGemma crashes for inputs longer than sliding window length (#37613)

manueldeprada committed 255 days ago

Verified 413f9bbf

Fix ValueError when eval_do_concat_batches=False with examples (#37621)

jeffhataws committed 255 days ago

Verified 964a1b6b

[tests] Stricter generate + compilation test -- no recompilations allowed (#37629)

gante committed 255 days ago

Verified 85665a42

[test] update `test_past_key_values_format` (#37614)

gante committed 255 days ago

Verified 362fa37d

Add test to ensure unknown exceptions reraising in utils/hub.py::cached_files() (#37651)

manueldeprada committed 255 days ago

Verified 1cd110c6

Support loading Gemma3 QAT GGUF models (#37649)

Isotr0py committed 255 days ago

Verified c69e2345

Restructure torchao quantization examples (#37592)

jerryzh168 committed 255 days ago

Verified 7eb1107c

[fix gemma] Set default value for output_attentions parameter in Gemma2 and Gemma… (#37633)

chenin-wang committed 255 days ago

Verified 006530d2

[fix] make legacy bnb code work (#37331)

cyr0930 committed 255 days ago

Verified 31ea547b

Fix Qwen2.5-Omni get_chunked_index chunking functionality (#37631)

imkero committed 255 days ago

Verified 5f791281

Refactor phi doc (#37583)

JihadHammoud02 committed 255 days ago

Verified fee11906

Update longformer.md (#37622)

JihadHammoud02 committed 255 days ago

Verified b2db54f6

fix link in kv_cache.md (#37652)

manueldeprada committed 255 days ago

Verified 2c60a442

Allow Exclusion of Input IDs from RepetitionPenaltyLogitsProcessor (#37625)

alex-jw-brooks committed 255 days ago

Verified a42ba80f

Remove torchvision requirement from AutoImageProcessor (#37457)

LysandreJik committed 255 days ago

Verified 10776034

[kernels] use original forward at compile time (#37604)

gante committed 255 days ago

Verified 1930e750

Fix InternVL attention when using qk_norm (38B and 78B) (#37620)

yonigozlan committed 257 days ago

Verified 6daa3eeb

chore: update model card for SigLIP (#37585)

saswatmeher committed 258 days ago

Verified 27a25bee

Fixing the example in generation strategy doc (#37598)

jeasinema committed 258 days ago

Verified e1f379bb

Deprecate modeling_utils.py classes (#37298)

qubvel committed 258 days ago

Verified 4f58fc9c

Add InternVL (2.5 MPO) (#35968)

yonigozlan committed 258 days ago

Verified a2450112

fix issue that some example with no trainer use accelerator.end_train… (#37435)

we1559 committed 258 days ago

Verified b0c6ff5e

fix 2 encoder_decoder issues on XPU (#37572)

yao-matrix committed 258 days ago

Verified 6f5014ac

[VLMs] use only `xxx_token_id` for multimodal tokens (#37573)

zucchini-nlp committed 258 days ago

Verified 2ba6b92a

Model debugger upgrades (#37391)

molbap committed 258 days ago

Verified 4afd3f48

Newer Older