huggingface/transformers

Pull Requests Commits

Fix the fsdp config cannot work issue. (#37549)

yuanwu2017 committed 250 days ago

Verified a41b6d9b

Gemma3 is Torch Exportable (#37728)

guangy10 committed 250 days ago

Verified 816b3701

Fix error message in `hub.py` (#37796)

srai9 committed 252 days ago

Verified 397a5ede

fix performance issue in convert_ids_to_tokens (#37773)

martin-harmonic committed 252 days ago

Verified 6ce675ee

chore: update SigLIP2 model card (#37624)

saswatmeher committed 252 days ago

Verified 57c620bf

[i18n-KO] Translated `keypoint_detection.md` to Korean (#36649)

rlaalsrl0922 committed 252 days ago

Verified eb4afdd1

fix mpt test of different outputs from cuda (#37691)

jiqing-feng committed 253 days ago

Verified 555693fb

Force torch>=2.6 with torch.load to avoid vulnerability issue (#37785)

Cyrilvallez committed 253 days ago

Verified 0cfbf9c9

Fix tensor parallel with non-floating dtypes (#37790)

Cyrilvallez committed 253 days ago

Verified eefc86aa

Fix typos in strings and comments (#37784)

co63oc committed 253 days ago

Verified 21406220

Align gpt2 mask preparation to #37612 (#37787)

Cyrilvallez committed 253 days ago

Verified ba3bd372

unpin pytest<8 (#37768)

ydshieh committed 253 days ago

Verified 50d231a8

[causal mask] fix preparation with multi-gpu (#37612)

zucchini-nlp committed 253 days ago

Verified 79d4bc76

🌐 [i18n-KO] Translated `roberta.md` to Korean (#37069)

garongkim committed 254 days ago

Verified 7bb619d7

Update model card for Gemma (#37674)

afafelwafi committed 254 days ago

Verified cfe66691

Fix auto-round hfoption (#37759)

MekkCyber committed 254 days ago

Verified b2d70e9c

Guard DeepSpeed imports (#37755)

lewtun committed 254 days ago

Verified acdbe627

[deps] pin max `torch` version (#37760)

gante committed 254 days ago

Verified af6d2756

Fix typos in comments (#37694)

co63oc committed 254 days ago

Verified 0302aa1c

Fix load of rng state for resuming training from checkpoint (#37162)

winglian committed 254 days ago

Verified af000ceb

Fix tied weight loading with TP and loading sub state_dicts (#37758)

Cyrilvallez committed 254 days ago

Verified 0af0a5f9

Refine parameter type annotations (#37666)

flashJd committed 254 days ago

Verified 3af24f7e

Fix wrong input shapes in doc-string of models (#37729)

kkew3 committed 254 days ago

Verified 22e3da92

[generate] fix default autocompile case on gpu (#37756)

gante committed 254 days ago

Verified 4d64c385

Fix qwen2_5 get_rope_index tensor device locations (#37597)

rphmeier committed 254 days ago

Verified 43bb4c04

updated hidden_features for FlaxDinov2SwiGLUFFN in Dinov2 (#37747)

premmurugan229 committed 254 days ago

Verified dd2649fa

[generate] skip compilation on cpu offload (#37709)

gante committed 254 days ago

Verified 8bdd4f2a

`GPT2Model` StaticCache support (#35761)

poedator committed 254 days ago

Verified 7c62e693

[cache] fix `HybridCache` init when `device` is passed (#37718)

gante committed 254 days ago

Verified 9f927c82

Expand quantized data type support for tensor parallelism (#37719)

amd-xiaoyu12 committed 254 days ago

Verified 4fee3209

Newer Older