huggingface/transformers

Pull Requests Commits

fix(pipelines): respect audio classification problem type (#50329)

BlazeSR10hawk committed 5 days ago

Verified cc405bb3

fix: forward tool schemas to response parsers (#50385)

BlazeSR10hawk committed 5 days ago

Verified de058827

Doge: note that the MoE router split is not causal in a full forward pass (#50420)

Andreas-Hum committed 5 days ago

Verified 9c5bdb1d

Update the longrope inv_freq buffer in place (#50480)

hmellor committed 5 days ago

Verified b2556881

Retry the attention kernel load after a failed `lazy_import_flash_attention` (#50391)

albertvillanova committed 5 days ago

Verified 75ca9731

Keep the GLM-5.3-Flash mHC base and scale in float32 (#50413)

albertvillanova committed 5 days ago

Verified b6401d66

fix failed cases for beam model (#50235)

kaixuanliu committed 5 days ago

Verified bbbf45fb

[torch 2.15] Update qwen3_vl_moe FA2 integration test expected values (#50481)

ydshieh committed 5 days ago

Verified f262a8cd

fix(gemma3n): Clamp the AltUp magnitudes instead of using a CPU epsilon tensor (#50457)

harshaljanjani committed 5 days ago

Verified 431e9e9d

[Mistral Large 4] Fix flaky test_left_padding_compatibility (#50479)

ydshieh committed 6 days ago

Verified de35c2b1

[torch 2.15] Update integration test expected values for A10 (cuda 8.6) (#50478)

ydshieh committed 6 days ago

Verified 0f7a8768

[Docker] Update Nvidia CI and Flash Attn CI images to torch 2.15 / cu132 (#50467)

ydshieh committed 7 days ago

Verified a6f5c867

Clarify warmup_steps error message to warmup_steps >= 0 (#50364)

adi274903 committed 8 days ago

Verified 0f53828e

[Gemma4] Stop prefill at last KV-owning layer when logits_to_keep=1 (#49640)

pculliton committed 8 days ago

Verified 2c6a00f9

Fix duplicated docstring (#50446)

Cyrilvallez committed 8 days ago

Verified 073a202c

chore(ci): refresh workflow references (#50436) (#50437)

hf-security-analysis[bot] committed 8 days ago

Verified 04b0d424

[serge] Fix 2 integration tests for model `pegasus` failing with `other` (other (2)) (#49507)

sergereview[bot] committed 8 days ago

Verified d9246b06

[distributed - TPU] Fix tied-weight comparisons on TPUs when using TP (#50406)

tengomucho committed 8 days ago

Verified 92b21f8d

vLLM CI: fix install and KV-cache memory, add failure report job (#50436)

ydshieh committed 8 days ago

Verified 826edb70

Harden our Jinja sandboxing (#50431)

Rocketknight1 committed 8 days ago

Verified 316217d3

Weight loading: conversion ops declare many-to-many support, and four small loader fixes (#49773)

IlyasMoutawwakil committed 8 days ago

Verified cea26862

Fix DeepSeek-V4 compression cache reordering for beam search (#50348)

taking-lying-flat committed 8 days ago

Verified 22c26e79

Allow saving an unchanged generation config (#49914)

benthecarman committed 8 days ago

Verified 7fb5f73b

[loading] Fix mismatched keys being silently ignored (#50433)

Cyrilvallez committed 8 days ago

Verified b6bb35a3

Capture only the projected hidden states in VLM feature extraction (#49594)

hmellor committed 8 days ago

Verified c5bd76ec

[Mistral Large 4] Fix flaky test_cached_decode_matches_cacheless (#50435)

ydshieh committed 8 days ago

Verified 96cff524

[`Colpali`] Fix offload tests (#50432)

vasqu committed 8 days ago

Verified d0efef1e

zucchini-nlp committed 8 days ago

Verified c6329dd0

OpenVINO exporter: let OpenVINO fuse attention, RoPE, the KV cache and compressed matmuls (#50311)

IlyasMoutawwakil committed 8 days ago

Verified 72b487d1

[VibeVoice] Avoid GPU syncs when removing audio padding (#50417)

hmellor committed 8 days ago

Verified 9fa60240

Newer Older