huggingface/transformers

Pull Requests Commits

Merge branch 'main' into gpt-oss-experts-impl

IlyasMoutawwakil committed 3 hours ago

Verified aa34996f

Simplify `_get_resolved_checkpoint_files` and do not try safetensors bot conversion with a subfolder (#43301)

Cyrilvallez committed 3 hours ago

Verified d12672fc

no need for logger

IlyasMoutawwakil committed 4 hours ago

4b0323ce

avoid baddbmm as it is less performant / less optimizable by max-autotune

IlyasMoutawwakil committed 4 hours ago

1c491124

quantization conversions should be applied first

IlyasMoutawwakil committed 4 hours ago

52e07786

IlyasMoutawwakil committed 5 hours ago

dd377e19

fix subfolder issue (#43300)

ArthurZucker committed 5 hours ago

Verified a85d6b2a

force 16 bytes alignmenet during weight loading

IlyasMoutawwakil committed 7 hours ago

2f3fd11c

[Qwen-VL] min/max pixels gone forever (#43279)

zucchini-nlp committed 8 hours ago

Verified 2b13b08d

IlyasMoutawwakil committed 8 hours ago

9dbed89b

[docs] remove old docs (#43156)

stevhliu committed 21 hours ago

Verified 7be2d358

Fix pipeline tests (#43293)

Cyrilvallez committed 22 hours ago

Verified 405d18df

[CB] Refactor: extract common scheduling logic into Scheduler base class (#43237)

dstekanov committed 22 hours ago

Verified 3af7c35f

Fix failing `fast_vlm` tests (#43275)

Sai-Suraj-27 committed 23 hours ago

Verified 0a5d574d

Delete Text2TextPipeline, SummarizationPipeline and TranslationPipeline (#43256)

Rocketknight1 committed 23 hours ago

Verified fc532243

[`Docs`] Minimax M2 add release date (#43290)

vasqu committed 23 hours ago

Verified 34691a98

[ci] Fixing some failing tests for important models (#43231)

Abdennacer-Badaoui committed 1 day ago

Verified d1808f2c

Add LightOnOCR model implementation (#41621)

baptiste-aubertin committed 1 day ago

Verified 3d276453

Update src/transformers/models/gpt_oss/modeling_gpt_oss.py

IlyasMoutawwakil committed 1 day ago

Verified b1a71a79

fix crash in when running FSDP2+TP (#43226)

sywangyi committed 1 day ago

Verified 77146cc9

[CB] Ensure parallel decoding test passes using FA (#43277)

remi-or committed 1 day ago

Verified 61317f5a

Fix failing `PegasusX`, `Mvp` & `LED` model integration tests (#43245)

Sai-Suraj-27 committed 1 day ago

Verified 1efe1a63

fallback to eager

IlyasMoutawwakil committed 1 day ago

846adcad

no need to relaod

IlyasMoutawwakil committed 1 day ago

5bd25c75

[consistency] Ensure models are added to the `_toctree.yml` (#43264)

Cyrilvallez committed 1 day ago

Verified e8ae3731

[docs] tensorrt-llm (#43176)

stevhliu committed 1 day ago

Verified c85be989

[style] Fix init isort and align makefile and CI (#43260)

Cyrilvallez committed 1 day ago

Verified 38022fd8

don't rely on weight shapes as they can be square matrices

IlyasMoutawwakil committed 2 days ago

0261a467

IlyasMoutawwakil committed 2 days ago

e1dba4d3

revert transposition and simply support transposed experts to avoid modifying eager

IlyasMoutawwakil committed 2 days ago

be08fe48

Older