Pull Requests huggingface/transformers

Fixed EfficientNet interpolation method to match the original implementation

#42860 opened 2025-12-13 17:46 by ammarjmahmood

fix: Ensure device consistency in OneFormerProcessor outputs Fixes #42722

#42858 opened 2025-12-13 14:55 by leejianwoo-collab

Fix: dtype-safe weight initialization for quantized models (skip non-floating tensors)

#42857 opened 2025-12-13 13:41 by ladder2sky

Improve tokenizer_type error message clarity (Issue #41553)

#42856 opened 2025-12-13 06:38 by alivezvaei110-lab

This lie is just to chck

#42855 opened 2025-12-13 04:47 by Anjanapramod281

Qwen2/3 MoE + GGUF model support (restored)

#42854 opened 2025-12-13 00:04 by a4lg

[docs] serving

#42853 opened 2025-12-12 22:44 by stevhliu

remove ipex and ccl for xpu and cpu

#42852 opened 2025-12-12 22:18 by yao-matrix

Add `.on_push_begin()` callback to Trainer and implement for TrackioCallback

#42850 opened 2025-12-12 21:00 by abidlabs

Fix: Improve AutoTokenizer error message for Voxtral model

#42849 opened 2025-12-12 19:37 by tamiraji-bot

[`Attn`] More new interface switches and proper paddingfree test

#42848 opened 2025-12-12 19:11 by vasqu

Remove `requests` code

#42845 opened 2025-12-12 17:57 by omkar-334

[modular] Fix a weird renaming edge-case

#42844 opened 2025-12-12 17:41 by Cyrilvallez

Pass in position_ids into attention for GPT2

#42842 opened 2025-12-12 16:03 by jackzhxng

Make SigLip implementation less weird

#42841 opened 2025-12-12 16:02 by zucchini-nlp

Remove tied weight keys Sam2Video

#42840 opened 2025-12-12 14:59 by yonigozlan

[CB] Easy optimizations for continuous batching

#42839 opened 2025-12-12 14:48 by remi-or

Fix speccht5_tts pipeline

#42830 opened 2025-12-12 05:30 by jiqing-feng

[WIP] End-to-end exportable pipelines (object detection)

#42829 opened 2025-12-12 05:21 by yonigozlan

[docs] optimization cleanup

#42827 opened 2025-12-12 00:44 by stevhliu

Fix torch only support for fast Processors

#42824 opened 2025-12-11 22:46 by Blaizzy

Add an integration test for LASR using pipe and chunked decoding

#42823 opened 2025-12-11 20:37 by kho

Rewrite for loop in get_image_features with torch ops for export

#42822 opened 2025-12-11 20:22 by jackzhxng

[kernels] Centralize kernels tests

#42819 opened 2025-12-11 16:20 by MekkCyber

Update `param_element_size`

#42818 opened 2025-12-11 15:38 by SunMarc

validate tokenizer components

#42816 opened 2025-12-11 15:31 by itazap

Single config attribute for weight tying

#42815 opened 2025-12-11 15:11 by zucchini-nlp

add tensor parallel test for MoE

#42809 opened 2025-12-11 14:04 by 3outeille

Remove duplicated processor class from config

#42806 opened 2025-12-11 10:47 by zucchini-nlp

[Fix] Fix FA2 kernels ut

#42803 opened 2025-12-11 10:02 by YangKai0616