huggingface/transformers

Pull Requests Commits

IlyasMoutawwakil committed 23 hours ago

1b7f9a0a

fix routers that returned the entire distribution of experts

IlyasMoutawwakil committed 23 hours ago

4a435530

Patch set_attn_implementation for Windows (#43433)

tomaarsen committed 1 day ago

Verified 388313e7

Make bot don't fail the commands (#43440)

ydshieh committed 1 day ago

Verified e4f63302

Remove `requests` code (#42845)

omkar-334 committed 1 day ago

Verified db1e6f1e

[`BC`] Update `get_(text|image|audio|video)_features` methods to return `BaseModelOutputWithPooling` (#42564)

tomaarsen committed 1 day ago

Verified 55dadb86

Yet another fix (#43437)

ydshieh committed 1 day ago

Verified c173472e

fix `pr-repo-consistency-bot.yml` (#43434)

ydshieh committed 1 day ago

Verified 236a8e9d

Make `pr-repo-consistency-bot.yml` run complete list of fixes (#43432)

ydshieh committed 1 day ago

Verified a1306676

Chore(tests): Fix values for test_inference_for_video_classification (#43406)

tarekziade committed 1 day ago

Verified f60129d1

Fix in-place modification of inputs_embeds in Kosmos-2.5 forward (#43418)

Neuro1729 committed 1 day ago

Verified 95be7ef5

Fix Mamba2ForCausalLM weight tying (#43207)

Anri-Lombard committed 1 day ago

Verified 10e97cd5

🚨Fix default interpolation for PVT to match original implementation (Bicubic) (#43017)

Arpit2744 committed 1 day ago

Verified ef2de73b

Fix tokenizer auto_map being ignored for custom models (#43219)

Anri-Lombard committed 1 day ago

Verified 54104656

SAM2 Video support fp16 (#43268)

Guppy16 committed 1 day ago

Verified 4a1ad8da

Fix: adding pad_token_id in Qwen3VLTextConfig (#43398)

vaibhav-research committed 1 day ago

Verified 62236a0a

Fix failing `recurrent_gemma` tests (#43405)

Sai-Suraj-27 committed 1 day ago

Verified 1f1532ad

Fix failing `ChameleonIntegrationTests` (#43401)

Sai-Suraj-27 committed 1 day ago

Verified 1908fb61

Fix typo in docstring for `Sam3TrackerConfig` (#43368)

alvarobartt committed 1 day ago

Verified 5c77298c

Fix always issuing generation deprecation warning (#43389)

githubnemo committed 1 day ago

Verified 6e1cde67

Fix typo in custom models documentation (#43399)

anthonyylee committed 1 day ago

Verified 79a58a74

fix: access `rope_theta` correctly in `Llama4VisionRotaryEmbedding` (#43281)

Tcc0403 committed 2 days ago

Verified 199c0d0a

GptOss experts implementation (#43227)

IlyasMoutawwakil committed 2 days ago

Verified 2d4d8fe4

Make Transformers more torch-exportable and dynamo-friendly (#42317)

IlyasMoutawwakil committed 2 days ago

Verified eff263cd

Support any capturing groups in WeightTransform reverse mapping (#43205)

yonigozlan committed 2 days ago

Verified de306e8e

Change Torchvision functional alias to TVF (#43263)

yonigozlan committed 2 days ago

Verified 9a6e8a2d

Add Solar-Open Model (#43244)

oesni committed 2 days ago

Verified 93dd4fb9

Fix non-deterministic output in MoE experts with batched_mm/grouped_mm (#43362)

raimbekovm committed 2 days ago

Verified 16904622

🚨Default to fast image processors for all models (#41388)

yonigozlan committed 2 days ago

Verified 3fec8c25

Add conversion mapping to Qwen3 Omni MoE Thinker submodel (#43373)

IlyasMoutawwakil committed 2 days ago

Verified eebf8562

Older