huggingface/transformers

Pull Requests Commits

Merge branch 'main' into remove-slice

Cyrilvallez committed 24 days ago

Verified e3bf2091

[loading] Really initialize on meta device for huge perf gains (#42941)

Cyrilvallez committed 24 days ago

Verified bb9357ff

fix error: 'BlockMask' object has no attribute 'dtype' for lasr model (#42802)

kaixuanliu committed 25 days ago

Verified 789226c1

fix concat order (#42946)

eustlb committed 25 days ago

Verified a4d62291

[kernels] Fix failling tests (#42953)

MekkCyber committed 25 days ago

Verified cf0f071e

[Quantization] CI green by end of year (#42951)

MekkCyber committed 25 days ago

Verified b5eea347

Updated `backbone_config` docstrings and type annotations (#42927)

simonreise committed 25 days ago

Verified 3e4baf8e

[docs] dtype (#42883)

stevhliu committed 25 days ago

Verified b93f2e3a

Load generation config from nested configs (#42922)

zucchini-nlp committed 25 days ago

Verified f2c6d2ad

🚨 Fix ConvNeXt image processor default interpolation to BICUBIC (#42934)

lukepayyapilli committed 25 days ago

Verified 60634caa

fix(tvp): add missing type_vocab_size parameter to TvpConfig (#42928)

majiayu000 committed 25 days ago

Verified b62e5b3e

[Quantization] Removing misleading int8 quantization in Finegrained FP8 (#42945)

MekkCyber committed 25 days ago

Verified 4dc60fd3

Fp8 dq (#42926)

SunMarc committed 25 days ago

Verified af91c0ba

Add buffers to `_init_weights` for ALL models (#42309)

Cyrilvallez committed 25 days ago

Verified 537c2e3d

rewrite _process_parameter_type in auto_docstring.py to improve usability (#42431)

Yacklin committed 25 days ago

Verified 728f34c3

Update `param_element_size` (#42818)

SunMarc committed 25 days ago

Verified dd8057af

[Tests] Fix CompressedTensors tests (#42935)

kylesayrs committed 26 days ago

Verified 0a846542

[CB] Allow block sharing in hybrid models (#42877)

remi-or committed 26 days ago

Verified 04e78e67

[Quantization] Misc tests fixes (#42940)

MekkCyber committed 26 days ago

Verified d7dd443a

[Quantization] rm _pre_quantization_dtype from quantization tests (#42939)

MekkCyber committed 26 days ago

Verified d3d4b629

🚨 Generation config defaults are now `None` (#42702)

zucchini-nlp committed 26 days ago

Verified a81e04a9

Document new default shard size + dropped unsafe serialization (#42904)

Wauplin committed 26 days ago

Verified bdaddb6f

Improve BatchFeature (.to() works on lists/nested lists of tensors, auto-skip non array conversion) (#42884)

yonigozlan committed 26 days ago

Verified 9f583b1b

[docs] optimization cleanup (#42827)

stevhliu committed 26 days ago

Verified 5ef16edd

[docs] Update shard size (#42749)

stevhliu committed 26 days ago

Verified 12fe95f8

[docs] Expert parallelism (#42409)

stevhliu committed 26 days ago

Verified 9e3568e0

[docs] WeightConverter (#42636)

stevhliu committed 26 days ago

Verified 1aab1e9c

docs: Squared ReLU paper fix (#42931)

casinca committed 26 days ago

Verified 2f9e21f5

Remove ipex/ccl in CPU training doc (#42866)

jiqing-feng committed 26 days ago

Verified f404f150

Make gradient-checkpoint enabling tolerant of models without get_input_embeddings (#42558)

molbap committed 26 days ago

Verified b712a97d

Older