huggingface/transformers

Pull Requests Commits

[WIP] add deepseek-v3 (#35926)

bzantium committed 1 year ago

Verified eca74d13

[blip-2] Fix dtype mismatch when keep in fp32 (#37068)

zucchini-nlp committed 1 year ago

Verified 52cc204d

Change deprecated PT functions (#37041)

cyyever committed 1 year ago

Verified aa3778af

Fix some typos about benchmark scripts. (#37027)

zhanluxianshen committed 1 year ago

Verified c90e6e96

Use `lru_cache` for tokenization tests (#36818)

ydshieh committed 1 year ago

Verified 1fcaad6d

fix: AttributeError: 'LlavaProcessor' object has no attribute 'image_token_id' (#37026)

jp1924 committed 1 year ago

Verified 3af425d4

Fix SDPA implementation in Qwen2-VL (issues with torch==2.6.0) (#36891)

ManuelFay committed 1 year ago

Verified 064cd7cd

fix: Fully remove legacy cache from Llama (#36958)

Wheest committed 1 year ago

Verified 348f3285

fixed typo (#37036)

finnoh committed 1 year ago

Verified d6b3c748

Remove deprecated batch_size parameter (#37007)

cyyever committed 1 year ago

Verified 6cc9c8d7

Replace default split function with jnp.split() in flax models (#37001)

premmurugan229 committed 1 year ago

Verified 4cc65e99

Set weights_only in torch.load (#36991)

cyyever committed 1 year ago

Verified 41a0e58e

Fix typing for None valued variables (#37004)

cyyever committed 1 year ago

Verified de77f5b1

Avoid unnecessary device operations in loss computing (#36950)

cyyever committed 1 year ago

Verified 8c5e29ba

clean pipeline question_answering. (#36986)

zhanluxianshen committed 1 year ago

Verified 471cf1de

[generate, cache] handle more complex device maps (#37014)

gante committed 1 year ago

Verified 29f322d0

[audio utils] fix fft_bin_width computation (#36603)

eustlb committed 1 year ago

Verified fb8e6c50

[chat templates} support loading audio from video (#36955)

zucchini-nlp committed 1 year ago

Verified e97c7600

Fixup for distill_any_depth conversion script (#37043)

qubvel committed 1 year ago

Verified c7bc79bd

Optimize `to_py_obj` for python-native numeric lists and scalars (#36885)

n0gu-furiosa committed 1 year ago

Verified d1eafe8d

fix pegasus init weights and other copied models (#36844)

jiqing-feng committed 1 year ago

Verified 0e56fb69

Add Distill Any Depth (#36614)

keetrap committed 1 year ago

Verified 7e813f9c

Skip FP8 linear tests For device capability < 9.0(#37008)

MekkCyber committed 1 year ago

Verified 92429057

remove redundant code in trainer (#36994)

hiyouga committed 1 year ago

Verified 279c2e30

Mark 2 tests as flaky for now (#37038)

ydshieh committed 1 year ago

Verified d13c390d

[Modeling] Load FP8 safetensors such as DeepSeek (#36828)

kylesayrs committed 1 year ago

Verified d6d930a6

Fix PixtralProcessor patch_size when spatial_merge_size is used (#37019)

mgoin committed 1 year ago

Verified 927ce1d3

Support QuestionAnswering Module for ModernBert based models. (#35566)

bakrianoo committed 1 year ago

Verified 49b5ab6a

fix transformers_cli import relative path issue (#36989)

yao-matrix committed 1 year ago

Verified 5b08db88

[docs] Attention mask image (#36970)

stevhliu committed 1 year ago

Verified 3a8ec8c4

Newer Older