Pull Requests huggingface/transformers

fix gemma4 has flash-attention incompatbile head-dim=512

#45202 opened 2026-04-02 23:37 by Qubitium

fix(models): Resolve regressions in Wav2Vec2PhonemeCTCTokenizer (wav2vec2-lv-60-espeak-cv-ft)

#45199 opened 2026-04-02 20:03 by harshaljanjani

Use torchvision `decode_image` to load images in the torchvision backend

#45195 opened 2026-04-02 17:58 by yonigozlan

Configuration insoncistencies

#45194 opened 2026-04-02 16:39 by zucchini-nlp

Config can apply pyndatic validation without torch-dependence

#45193 opened 2026-04-02 15:36 by zucchini-nlp

Add edge case tests for out-of-range token id decoding in Qwen2 tokenizer

#45191 opened 2026-04-02 14:08 by saslifat-gif

Fix ty for transformers cli

#45190 opened 2026-04-02 13:54 by SunMarc

Add doc test CI workflow reusing existing model job infrastructure

#45189 opened 2026-04-02 13:43 by ydshieh

Close file handler

#45187 opened 2026-04-02 13:10 by ydshieh

Add new model: Isaac

#45186 opened 2026-04-02 12:29 by zucchini-nlp

[CB] [Major] Add CPU request offloading

#45184 opened 2026-04-02 10:12 by remi-or

Make the cli a top-level package

#45181 opened 2026-04-02 08:03 by tarekziade

added efficietvitsam model to HF

#45176 opened 2026-04-02 00:47 by masoudpz

Fix Sam3Processor missing input_boxes_labels for padded None entries

#45171 opened 2026-04-01 18:26 by Kash6

`layrnorm` -> `layernorm`

#45170 opened 2026-04-01 17:50 by zucchini-nlp

Update min_lr and max_lr default values to better defaults

#45168 opened 2026-04-01 15:02 by w601sxs

Add anthropic style of function schema

#45167 opened 2026-04-01 14:50 by xu-song

Fix missing image processors backends

#45165 opened 2026-04-01 14:18 by zucchini-nlp

tweak checkers output on errors

#45163 opened 2026-04-01 13:41 by tarekziade

[WIP] PrismML Bonsai model support

#45157 opened 2026-04-01 03:11 by Qubitium

feat: adapter loading with TP

#45155 opened 2026-03-31 22:23 by michaelbenayoun

[`FA`] Native torch integration

#45153 opened 2026-03-31 19:43 by vasqu

[docs] model testing

#45152 opened 2026-03-31 18:12 by stevhliu

DO NOT MERGE adding SAML3-LiteText with a skill, first pass

#45149 opened 2026-03-31 15:56 by tarekziade

Fix broken HQQ support

#45147 opened 2026-03-31 14:59 by mobicham

Add Xiaomi MiMo-V2 New model

#45144 opened 2026-03-31 13:53 by casinca

Fix vllm cis

#45139 opened 2026-03-31 09:26 by ArthurZucker

Optimize Parakeet feature extraction on CUDA

#45134 opened 2026-03-31 01:59 by milesial

Add sarvam model

#45133 opened 2026-03-31 00:00 by itazap

Fix MoE routers returning probabilities instead of logits

#45131 opened 2026-03-30 21:18 by yacinemebarki