diff --git "a/data.js" "b/data.js" --- "a/data.js" +++ "b/data.js" @@ -2,7 +2,10 @@ // Item spans are REAL: first commit of the earliest PR -> last commit / merge of the latest PR. // Source board: https://miro.com/app/board/uXjVL_vDn1k=/ -const AXIS = ["2024-06-04", "2026-08-26"]; +const AXIS = ["2024-06-04", "2026-08-31"]; +const RELEASES = [{"n": "v4.42.0", "d": "2024-06-27", "big": false}, {"n": "v4.43.0", "d": "2024-07-23", "big": false}, {"n": "v4.44.0", "d": "2024-08-06", "big": false}, {"n": "v4.45.0", "d": "2024-09-25", "big": false}, {"n": "v4.46.0", "d": "2024-10-24", "big": false}, {"n": "v4.47.0", "d": "2024-12-05", "big": false}, {"n": "v4.48.0", "d": "2025-01-10", "big": false}, {"n": "v4.49.0", "d": "2025-02-17", "big": false}, {"n": "v4.50.0", "d": "2025-03-21", "big": false}, {"n": "v4.51.0", "d": "2025-04-05", "big": false}, {"n": "v4.52.0", "d": "2025-05-20", "big": false}, {"n": "v4.53.0", "d": "2025-06-26", "big": false}, {"n": "v4.54.0", "d": "2025-07-25", "big": false}, {"n": "v4.55.0", "d": "2025-08-05", "big": false}, {"n": "v4.56.0", "d": "2025-08-29", "big": false}, {"n": "v4.57.0", "d": "2025-10-03", "big": false}, {"n": "v5.0.0rc0", "d": "2025-12-01", "big": true}, {"n": "v5.0.0rc1", "d": "2025-12-11", "big": false}, {"n": "v5.0.0rc2", "d": "2026-01-07", "big": false}, {"n": "v5.0.0rc3", "d": "2026-01-14", "big": false}, {"n": "v5.0.0", "d": "2026-01-26", "big": true}, {"n": "v5.1.0", "d": "2026-02-05", "big": false}, {"n": "v5.2.0", "d": "2026-02-16", "big": false}, {"n": "v5.3.0", "d": "2026-03-04", "big": false}, {"n": "v5.4.0", "d": "2026-03-27", "big": false}, {"n": "v5.5.0", "d": "2026-04-02", "big": false}, {"n": "v5.6.0", "d": "2026-04-22", "big": false}, {"n": "v5.7.0", "d": "2026-04-28", "big": false}, {"n": "v5.8.0", "d": "2026-05-05", "big": false}, {"n": "v5.9.0", "d": "2026-05-20", "big": false}, {"n": "v5.10.0", "d": "2026-06-03", "big": false}, {"n": "v5.11.0", "d": "2026-06-10", "big": false}, {"n": "v5.12.0", "d": "2026-06-12", "big": false}, {"n": "v5.13.0", "d": "2026-07-03", "big": false}, {"n": "v5.14.0", "d": "2026-07-15", "big": false}, {"n": "v5.15.0", "d": "2026-08-10", "big": false}, {"n": "v5.16.0", "d": "2026-08-26", "big": false}]; +const MODELS = [{"m": "rt_detr", "d": "2024-06-22", "pr": 29077, "by": "Sangbum Daniel Choi", "t": "New model support RTDETR"}, {"m": "instructblipvideo", "d": "2024-06-25", "pr": 30182, "by": "Raushan Turganbay", "t": "Add video modality for InstrucBLIP"}, {"m": "llava_next_video", "d": "2024-06-26", "pr": 31252, "by": "Raushan Turganbay", "t": "Add LLaVa NeXT Video"}, {"m": "gemma2", "d": "2024-06-27", "pr": 31659, "by": "Arthur", "t": "Add gemma 2"}, {"m": "zoedepth", "d": "2024-07-08", "pr": 30136, "by": "NielsRogge", "t": "Add ZoeDepth"}, {"m": "hiera", "d": "2024-07-12", "pr": 30356, "by": "Naman Garg", "t": "Adding hiera"}, {"m": "chameleon", "d": "2024-07-17", "pr": 31534, "by": "Raushan Turganbay", "t": "Chameleon: add model"}, {"m": "nemotron", "d": "2024-08-06", "pr": 31699, "by": "Ao Tang", "t": "Add Nemotron HF Support"}, {"m": "mamba2", "d": "2024-08-06", "pr": 32080, "by": "Pablo Montalvo", "t": "Add codestral mamba2"}, {"m": "qwen2_audio", "d": "2024-08-08", "pr": 32137, "by": "Yunfei Chu", "t": "Add Qwen2-Audio"}, {"m": "falcon_mamba", "d": "2024-08-12", "pr": 32615, "by": "Younes Belkada", "t": "Add new model"}, {"m": "dac", "d": "2024-08-19", "pr": 31494, "by": "Kamil Akesbi", "t": "Add Descript-Audio-Codec model"}, {"m": "qwen2_vl", "d": "2024-08-26", "pr": 32318, "by": "Shijie", "t": "support qwen2-vl"}, {"m": "granite", "d": "2024-08-27", "pr": 31502, "by": "Mayank Mishra", "t": "Granite language models"}, {"m": "olmoe", "d": "2024-09-03", "pr": 32406, "by": "Niklas Muennighoff", "t": "Add OLMoE"}, {"m": "llava_onevision", "d": "2024-09-05", "pr": 32673, "by": "Raushan Turganbay", "t": "Llava Onevision: add model"}, {"m": "pixtral", "d": "2024-09-14", "pr": 33449, "by": "Arthur", "t": "Add support for Pixtral"}, {"m": "mimi", "d": "2024-09-18", "pr": 33565, "by": "Yoach Lacombe", "t": "Codec integration"}, {"m": "granitemoe", "d": "2024-09-20", "pr": 33207, "by": "Mayank Mishra", "t": "Granitemoe"}, {"m": "omdet_turbo", "d": "2024-09-25", "pr": 31843, "by": "Yoni Gozlan", "t": "Add OmDet-Turbo"}, {"m": "mllama", "d": "2024-09-25", "pr": 33703, "by": "Arthur", "t": "Add MLLama"}, {"m": "idefics3", "d": "2024-09-25", "pr": 32473, "by": "Andrés Marafioti", "t": "Add Idefics 3!"}, {"m": "phimoe", "d": "2024-10-04", "pr": 33363, "by": "Amit Garg", "t": "PhiMoE"}, {"m": "zamba", "d": "2024-10-04", "pr": 30950, "by": "pglorio", "t": "Add Zamba"}, {"m": "myt5", "d": "2024-10-06", "pr": 31286, "by": "TomLim", "t": "[WIP] Add Tokenizer for MyT5 Model"}, {"m": "moshi", "d": "2024-10-16", "pr": 33624, "by": "Yoach Lacombe", "t": "Moshi integration"}, {"m": "glm", "d": "2024-10-18", "pr": 33823, "by": "Cyril Vallez", "t": "add Glm"}, {"m": "olmo_1124", "d": "2024-11-18", "pr": 34551, "by": "Shane A", "t": "Add OLMo November 2024"}, {"m": "ijepa", "d": "2024-12-05", "pr": 33125, "by": "João Marcelo", "t": "Add I-JEPA"}, {"m": "aria", "d": "2024-12-06", "pr": 34157, "by": "Aymeric Roucher", "t": "Add Aria"}, {"m": "timm_wrapper", "d": "2024-12-11", "pr": 34564, "by": "Pavel Iakubovskii", "t": "Add TimmWrapper"}, {"m": "cohere2", "d": "2024-12-13", "pr": 35224, "by": "alexrs-cohere", "t": "Add Cohere2 model"}, {"m": "colpali", "d": "2024-12-17", "pr": 33736, "by": "Tony Wu", "t": "Add ColPali to 🤗 transformers"}, {"m": "bamba", "d": "2024-12-19", "pr": 34982, "by": "Yu Chin Fabian Lim", "t": "Add the Bamba Model"}, {"m": "modernbert", "d": "2024-12-19", "pr": 35158, "by": "Benjamin Warner", "t": "Add ModernBERT to Transformers"}, {"m": "dinov2_with_registers", "d": "2024-12-24", "pr": 35348, "by": "NielsRogge", "t": "Add DINOv2 with registers"}, {"m": "diffllama", "d": "2025-01-07", "pr": 34083, "by": "松本和真", "t": "Add diffllama"}, {"m": "textnet", "d": "2025-01-08", "pr": 34979, "by": "Jade Choghari", "t": "Add TextNet"}, {"m": "vitpose", "d": "2025-01-08", "pr": 30530, "by": "NielsRogge", "t": "Add ViTPose"}, {"m": "vitpose_backbone", "d": "2025-01-08", "pr": 30530, "by": "NielsRogge", "t": "Add ViTPose"}, {"m": "moonshine", "d": "2025-01-10", "pr": 34784, "by": "eustlb", "t": "Add Moonshine"}, {"m": "emu3", "d": "2025-01-10", "pr": 33770, "by": "Raushan Turganbay", "t": "[WIP] Emu3: add model"}, {"m": "helium", "d": "2025-01-13", "pr": 35669, "by": "Arthur", "t": "Add-helium"}, {"m": "superglue", "d": "2025-01-20", "pr": 29886, "by": "StevenBucaille", "t": "Add SuperGlue model"}, {"m": "qwen2_5_vl", "d": "2025-01-23", "pr": 35569, "by": "ShuaiBai623", "t": "add qwen2.5vl"}, {"m": "zamba2", "d": "2025-01-27", "pr": 34517, "by": "pglorio", "t": "Add Zamba2"}, {"m": "got_ocr2", "d": "2025-01-31", "pr": 34721, "by": "Yoni Gozlan", "t": "Add GOT-OCR 2.0 to Transformers"}, {"m": "dab_detr", "d": "2025-02-04", "pr": 30803, "by": "David", "t": "Add DAB-DETR for object detection"}, {"m": "rt_detr_v2", "d": "2025-02-06", "pr": 34773, "by": "Jade Choghari", "t": "Adding RT-DETRv2 for object detection"}, {"m": "depth_pro", "d": "2025-02-10", "pr": 34583, "by": "Armaghan Shakir", "t": "Add Apple's Depth-Pro for depth estimation"}, {"m": "granitemoeshared", "d": "2025-02-14", "pr": 35894, "by": "Mayank Mishra", "t": "add shared experts for upcoming Granite 4.0 language models"}, {"m": "smolvlm", "d": "2025-02-20", "pr": 36126, "by": "Orr Zohar", "t": "SmolVLM2"}, {"m": "siglip2", "d": "2025-02-21", "pr": 36323, "by": "Pavel Iakubovskii", "t": "Add SigLIP 2"}, {"m": "aya_vision", "d": "2025-03-04", "pr": 36521, "by": "Arthur", "t": "Add aya"}, {"m": "gemma3", "d": "2025-03-12", "pr": 36658, "by": "Ryan Mullins", "t": "Gemma3"}, {"m": "mistral3", "d": "2025-03-18", "pr": 36790, "by": "Cyril Vallez", "t": "Add Mistral3"}, {"m": "shieldgemma2", "d": "2025-03-20", "pr": 36678, "by": "Ryan Mullins", "t": "Shieldgemma2"}, {"m": "prompt_depth_anything", "d": "2025-03-21", "pr": 35401, "by": "Haotong LIN", "t": "Add Prompt Depth Anything Model"}, {"m": "phi4_multimodal", "d": "2025-03-25", "pr": 36939, "by": "Cyril Vallez", "t": "Add Phi4 multimodal"}, {"m": "deepseek_v3", "d": "2025-03-28", "pr": 35926, "by": "Minho Ryu", "t": "[WIP] add deepseek-v3"}, {"m": "qwen3", "d": "2025-03-31", "pr": 36878, "by": "Bo Zheng", "t": "Adding Qwen3 and Qwen3MoE"}, {"m": "qwen3_moe", "d": "2025-03-31", "pr": 36878, "by": "Bo Zheng", "t": "Adding Qwen3 and Qwen3MoE"}, {"m": "llama4", "d": "2025-04-05", "pr": 37307, "by": "Arthur", "t": "Add llama4"}, {"m": "glm4", "d": "2025-04-09", "pr": 37388, "by": "Arthur", "t": "Add glm4"}, {"m": "granite_speech", "d": "2025-04-11", "pr": 36801, "by": "Alex Brooks", "t": "Add Granite Speech Support"}, {"m": "qwen2_5_omni", "d": "2025-04-14", "pr": 36752, "by": "BakerBunker", "t": "Add Qwen2.5-Omni"}, {"m": "mlcd", "d": "2025-04-15", "pr": 36182, "by": "Huajie Tan", "t": "Add MLCD model"}, {"m": "timesfm", "d": "2025-04-16", "pr": 34082, "by": "Jinan Zhou", "t": "Add TimesFM Time Series Forecasting Model"}, {"m": "janus", "d": "2025-04-17", "pr": 36053, "by": "Yaswanth Gali", "t": "Add Janus model"}, {"m": "internvl", "d": "2025-04-18", "pr": 35968, "by": "Yoni Gozlan", "t": "Add InternVL (2.5 MPO)"}, {"m": "bitnet", "d": "2025-04-28", "pr": 37742, "by": "Mohamed Mekkouri", "t": "Add Bitnet model"}, {"m": "sam_hq", "d": "2025-04-28", "pr": 35147, "by": "sushmanth reddy", "t": "Samhq model addition"}, {"m": "d_fine", "d": "2025-04-29", "pr": 36261, "by": "Vladislav Bronzov", "t": "Add D-FINE Model into Transformers"}, {"m": "hgnet_v2", "d": "2025-04-29", "pr": 36261, "by": "Vladislav Bronzov", "t": "Add D-FINE Model into Transformers"}, {"m": "granitemoehybrid", "d": "2025-05-05", "pr": 37658, "by": "Sukriti Sharma", "t": "Add GraniteMoeHybrid support for 4.0"}, {"m": "csm", "d": "2025-05-07", "pr": 36719, "by": "eustlb", "t": "Add CSM model"}, {"m": "falcon_h1", "d": "2025-05-21", "pr": 38249, "by": "Younes Belkada", "t": "[MODEL] Add Falcon H1"}, {"m": "colqwen2", "d": "2025-06-02", "pr": 35778, "by": "Tony Wu", "t": "Add ColQwen2 to 🤗 transformers"}, {"m": "minimax", "d": "2025-06-04", "pr": 35831, "by": "Armaghan Shakir", "t": "Add support for MiniMax's MiniMax-Text-01"}, {"m": "vjepa2", "d": "2025-06-11", "pr": 38746, "by": "Pavel Iakubovskii", "t": "Add V-JEPA 2"}, {"m": "lightglue", "d": "2025-06-17", "pr": 31718, "by": "StevenBucaille", "t": "Add LightGlue model"}, {"m": "arcee", "d": "2025-06-24", "pr": 38621, "by": "Crystalcareai", "t": "Add Arcee model support"}, {"m": "stt", "d": "2025-06-24", "pr": 38909, "by": "eustlb", "t": "Add kyutai stt"}, {"m": "glm4v", "d": "2025-06-25", "pr": 38431, "by": "Yuxuan Zhang", "t": "GLM-4.1V Model support"}, {"m": "t5gemma", "d": "2025-06-25", "pr": 38332, "by": "Biao Zhang", "t": "Encoder-Decoder Gemma"}, {"m": "dots1", "d": "2025-06-25", "pr": 38143, "by": "redmoe-moutain", "t": "[Model] add dots1"}, {"m": "smollm3", "d": "2025-06-25", "pr": 38755, "by": "Anton Lozhkov", "t": "Add SmolLM3"}, {"m": "dia", "d": "2025-06-26", "pr": 38405, "by": "Jaeyong Sung", "t": "Add Dia model"}, {"m": "gemma3n", "d": "2025-06-26", "pr": 39059, "by": "Ryan Mullins", "t": "Gemma 3n"}, {"m": "eomt", "d": "2025-06-27", "pr": 37610, "by": "Yaswanth Gali", "t": "✨ Add EoMT Model || 🚨 Fix Mask2Former loss calculation"}, {"m": "doge", "d": "2025-07-08", "pr": 35891, "by": "Jingze Shi", "t": "Add Doge model"}, {"m": "aimv2", "d": "2025-07-08", "pr": 36625, "by": "Yaswanth Gali", "t": "Add Aimv2 model"}, {"m": "deepseek_v2", "d": "2025-07-09", "pr": 36400, "by": "Vladislav Bronzov", "t": "Add DeepSeek V2 Model into Transformers"}, {"m": "lfm2", "d": "2025-07-10", "pr": 39340, "by": "Paul Pak", "t": "LFM2"}, {"m": "perception_lm", "d": "2025-07-11", "pr": 37878, "by": "Shuming Hu", "t": "PerceptionLM"}, {"m": "modernbert_decoder", "d": "2025-07-15", "pr": 38967, "by": "Orion Weller", "t": "Add ModernBERT Decoder Models - ModernBERT, but trained with CLM!"}, {"m": "voxtral", "d": "2025-07-18", "pr": 39429, "by": "eustlb", "t": "Add voxtral"}, {"m": "glm4_moe", "d": "2025-07-21", "pr": 39393, "by": "Yuxuan Zhang", "t": "GLM-4 Update"}, {"m": "ernie4_5", "d": "2025-07-21", "pr": 39228, "by": "Anton Vlasjuk", "t": "[`Ernie 4.5`] Add ernie text models"}, {"m": "ernie4_5_moe", "d": "2025-07-21", "pr": 39228, "by": "Anton Vlasjuk", "t": "[`Ernie 4.5`] Add ernie text models"}, {"m": "efficientloftr", "d": "2025-07-22", "pr": 36355, "by": "StevenBucaille", "t": "Add EfficientLoFTR model"}, {"m": "deepseek_vl", "d": "2025-07-25", "pr": 36248, "by": "Armaghan Shakir", "t": "Add support for DeepseekAI's DeepseekVL"}, {"m": "deepseek_vl_hybrid", "d": "2025-07-25", "pr": 36248, "by": "Armaghan Shakir", "t": "Add support for DeepseekAI's DeepseekVL"}, {"m": "xlstm", "d": "2025-07-25", "pr": 39665, "by": "Cyril Vallez", "t": "Add xlstm model"}, {"m": "evolla", "d": "2025-07-26", "pr": 36232, "by": "Xibin Bayes Zhou", "t": "Add evolla rebase main"}, {"m": "exaone4", "d": "2025-07-26", "pr": 39129, "by": "lgai-exaone", "t": "Add EXAONE 4.0 model"}, {"m": "cohere2_vision", "d": "2025-07-31", "pr": 39810, "by": "Raushan Turganbay", "t": "[Model] Cohere2 Vision"}, {"m": "mm_grounding_dino", "d": "2025-08-01", "pr": 37925, "by": "rziga", "t": "Add MM Grounding DINO"}, {"m": "gpt_oss", "d": "2025-08-05", "pr": 39923, "by": "Arthur", "t": "Add GPT OSS model from OpenAI"}, {"m": "glm4v_moe", "d": "2025-08-08", "pr": 39805, "by": "Yuxuan Zhang", "t": "GLM-4.5V Model Support"}, {"m": "sam2", "d": "2025-08-14", "pr": 32317, "by": "Sangbum Daniel Choi", "t": "Add Segment Anything 2 (SAM2)"}, {"m": "sam2_video", "d": "2025-08-14", "pr": 32317, "by": "Sangbum Daniel Choi", "t": "Add Segment Anything 2 (SAM2)"}, {"m": "dinov3_convnext", "d": "2025-08-14", "pr": 40167, "by": "Pavel Iakubovskii", "t": "Fix docs typo"}, {"m": "dinov3_vit", "d": "2025-08-14", "pr": 40167, "by": "Pavel Iakubovskii", "t": "Fix docs typo"}, {"m": "xcodec", "d": "2025-08-15", "pr": 38248, "by": "Manal ML", "t": "Add X-Codec model"}, {"m": "ovis2", "d": "2025-08-18", "pr": 37088, "by": "Eon Kim", "t": "Add Ovis2 model and processor implementation"}, {"m": "kosmos2_5", "d": "2025-08-19", "pr": 31711, "by": "tic-top", "t": "Add Kosmos-2.5"}, {"m": "metaclip_2", "d": "2025-08-20", "pr": 39826, "by": "NielsRogge", "t": "Add MetaCLIP 2"}, {"m": "florence2", "d": "2025-08-20", "pr": 38188, "by": "Duc-Viet Hoang", "t": "Add support for Florence-2"}, {"m": "hunyuan_v1_dense", "d": "2025-08-22", "pr": 39606, "by": "yjc9696", "t": "HunYuan opensource"}, {"m": "hunyuan_v1_moe", "d": "2025-08-22", "pr": 39606, "by": "yjc9696", "t": "HunYuan opensource"}, {"m": "seed_oss", "d": "2025-08-22", "pr": 40272, "by": "Fazzie", "t": "Addiing ByteDance Seed Seed-OSS"}, {"m": "apertus", "d": "2025-08-28", "pr": 39381, "by": "EduardDurech", "t": "Add Apertus"}, {"m": "qwen3_next", "d": "2025-09-10", "pr": 40771, "by": "Bo Zheng", "t": "Adding Support for Qwen3-Next"}, {"m": "ministral", "d": "2025-09-11", "pr": 40247, "by": "Manuel de Prada Corral", "t": "⚠️ 🔴 Add ministral model"}, {"m": "vaultgemma", "d": "2025-09-12", "pr": 40851, "by": "Ryan Mullins", "t": "add: differential privacy research model"}, {"m": "qwen3_vl", "d": "2025-09-15", "pr": 40795, "by": "JJJYmmm", "t": "Adding Support for Qwen3-VL Series"}, {"m": "qwen3_vl_moe", "d": "2025-09-15", "pr": 40795, "by": "JJJYmmm", "t": "Adding Support for Qwen3-VL Series"}, {"m": "olmo3", "d": "2025-09-16", "pr": 40778, "by": "Shane A", "t": "Add Olmo3 model"}, {"m": "longcat_flash", "d": "2025-09-17", "pr": 40730, "by": "Pablo Montalvo", "t": "Add LongCat-Flash"}, {"m": "flex_olmo", "d": "2025-09-18", "pr": 40921, "by": "Shane A", "t": "Add FlexOlmo model"}, {"m": "lfm2_vl", "d": "2025-09-18", "pr": 40624, "by": "Raushan Turganbay", "t": "Add new model LFM2-VL"}, {"m": "blt", "d": "2025-09-19", "pr": 38579, "by": "Ita Zaporozhets", "t": "blt wip"}, {"m": "qwen3_omni_moe", "d": "2025-09-21", "pr": 41025, "by": "BakerBunker", "t": "Adding support for Qwen3Omni"}, {"m": "parakeet", "d": "2025-09-25", "pr": 39062, "by": "Nithin Rao", "t": "Add Parakeet"}, {"m": "edgetam", "d": "2025-09-29", "pr": 39800, "by": "Yoni Gozlan", "t": "Add EdgeTAM"}, {"m": "edgetam_video", "d": "2025-09-29", "pr": 39800, "by": "Yoni Gozlan", "t": "Add EdgeTAM"}, {"m": "lfm2_moe", "d": "2025-10-07", "pr": 41401, "by": "Paul Pak", "t": "[Model] Lfm2Moe"}, {"m": "cwm", "d": "2025-10-09", "pr": 41199, "by": "Jacob Kahn", "t": "Add Code World Model (CWM)"}, {"m": "video_llama_3", "d": "2025-10-13", "pr": 40499, "by": "Kehan Li", "t": "[model] Add VideoLLaMA3 implementation"}, {"m": "audioflamingo3", "d": "2025-11-12", "pr": 40290, "by": "Lasha Koroshinadze", "t": "[models] Add AudioFlamingo3 integration"}, {"m": "glm46v", "d": "2025-11-15", "pr": 42122, "by": "Yuxuan Zhang", "t": "GLM-V update with new processor"}, {"m": "sam3", "d": "2025-11-19", "pr": 42285, "by": "Yoni Gozlan", "t": "Add SAM3 to 🤗 Transformers"}, {"m": "sam3_tracker", "d": "2025-11-19", "pr": 42285, "by": "Yoni Gozlan", "t": "Add SAM3 to 🤗 Transformers"}, {"m": "sam3_tracker_video", "d": "2025-11-19", "pr": 42285, "by": "Yoni Gozlan", "t": "Add SAM3 to 🤗 Transformers"}, {"m": "sam3_video", "d": "2025-11-19", "pr": 42285, "by": "Yoni Gozlan", "t": "Add SAM3 to 🤗 Transformers"}, {"m": "nanochat", "d": "2025-11-27", "pr": 41634, "by": "burtenshaw", "t": "[MODEL] Nanochat implementation"}, {"m": "afmoe", "d": "2025-11-29", "pr": 42168, "by": "Raghav Ravishankar", "t": "Add afmoe model"}, {"m": "t5gemma2", "d": "2025-12-01", "pr": 41834, "by": "Biao Zhang", "t": "T5gemma2"}, {"m": "ministral3", "d": "2025-12-01", "pr": 42498, "by": "Patrick von Platen", "t": "[Ministral 3] Add ministral 3"}, {"m": "fast_vlm", "d": "2025-12-02", "pr": 41112, "by": "Kamila Luczaj (Woo-chai)", "t": "Add FastVLM"}, {"m": "lasr", "d": "2025-12-05", "pr": 42648, "by": "eustlb", "t": "Lasr model"}, {"m": "paddleocr_vl", "d": "2025-12-11", "pr": 42178, "by": "zhang-prog", "t": "[Model] Add PaddleOCR-VL Model Support"}, {"m": "pe_audio", "d": "2025-12-16", "pr": 42905, "by": "eustlb", "t": " Sam: Perception Encoder Audiovisual"}, {"m": "pe_audio_video", "d": "2025-12-16", "pr": 42905, "by": "eustlb", "t": " Sam: Perception Encoder Audiovisual"}, {"m": "pe_video", "d": "2025-12-16", "pr": 42905, "by": "eustlb", "t": " Sam: Perception Encoder Audiovisual"}, {"m": "jais2", "d": "2025-12-17", "pr": 42684, "by": "sarathc-cerebras", "t": "adds jais2 model support"}, {"m": "pixio", "d": "2025-12-17", "pr": 42795, "by": "Lihe Yang", "t": "Add Pixio pre-trained models"}, {"m": "ernie4_5_vl_moe", "d": "2025-12-19", "pr": 39585, "by": "Anton Vlasjuk", "t": "[`Ernie 4.5`] Ernie VL models"}, {"m": "glmasr", "d": "2025-12-24", "pr": 42875, "by": "Yuxuan Zhang", "t": "GLM-ASR Support"}, {"m": "minimax_m2", "d": "2026-01-09", "pr": 42028, "by": "Roger Young", "t": "Add support for MiniMax-M2"}, {"m": "lw_detr", "d": "2026-01-12", "pr": 40991, "by": "StevenBucaille", "t": "Add LWDetr model"}, {"m": "glm4_moe_lite", "d": "2026-01-13", "pr": 43031, "by": "Yuxuan Zhang", "t": "[GLM-4.7] GLM-Lite Supoort"}, {"m": "glm_image", "d": "2026-01-13", "pr": 43100, "by": "Yuxuan Zhang", "t": "[GLM-Image] AR Model Support for GLM-Image"}, {"m": "lighton_ocr", "d": "2026-01-14", "pr": 41621, "by": "Aubertin Baptiste", "t": "Add LightOnOCR model implementation"}, {"m": "solar_open", "d": "2026-01-22", "pr": 43244, "by": "Inseo Song", "t": "Add Solar-Open Model"}, {"m": "glm_ocr", "d": "2026-01-27", "pr": 43391, "by": "Yuxuan Zhang", "t": "[GLM-OCR] GLM-OCR Support"}, {"m": "youtu", "d": "2026-01-28", "pr": 43166, "by": "Lu junru", "t": "Add Youtu-LLM model"}, {"m": "pp_doclayout_v3", "d": "2026-01-29", "pr": 43098, "by": "zhang-prog", "t": "[Model] Add PP-DocLayoutV3 Model Support"}, {"m": "eomt_dinov3", "d": "2026-02-02", "pr": 41212, "by": "NielsRogge", "t": "Add EoMT with DINOv3 backbone"}, {"m": "exaone_moe", "d": "2026-02-04", "pr": 43080, "by": "Junwon Hwang", "t": "Add EXAONE-MoE implementations"}, {"m": "moonshine_streaming", "d": "2026-02-04", "pr": 43702, "by": "eustlb", "t": "Add moonshine streaming"}, {"m": "vibevoice_acoustic_tokenizer", "d": "2026-02-06", "pr": 43400, "by": "Eric Bezzam", "t": "Add VibeVoice Acoustic Tokenizer"}, {"m": "qwen3_5", "d": "2026-02-09", "pr": 43830, "by": "Bo Zheng", "t": "Adding Support for Qwen3.5"}, {"m": "qwen3_5_moe", "d": "2026-02-09", "pr": 43830, "by": "Bo Zheng", "t": "Adding Support for Qwen3.5"}, {"m": "glm_moe_dsa", "d": "2026-02-09", "pr": 43858, "by": "Cyril Vallez", "t": "Add GlmMoeDsa"}, {"m": "voxtral_realtime", "d": "2026-02-16", "pr": 43769, "by": "eustlb", "t": "Add Voxtral Realtime"}, {"m": "higgs_audio_v2", "d": "2026-02-19", "pr": 40294, "by": "Shuai Zheng", "t": "Add Higgs Audio V2 Model"}, {"m": "higgs_audio_v2_tokenizer", "d": "2026-02-19", "pr": 40294, "by": "Shuai Zheng", "t": "Add Higgs Audio V2 Model"}, {"m": "colmodernvbert", "d": "2026-02-23", "pr": 42504, "by": "Paul Teiletche", "t": "Add ModernVBERT models"}, {"m": "modernvbert", "d": "2026-02-23", "pr": 42504, "by": "Paul Teiletche", "t": "Add ModernVBERT models"}, {"m": "olmo_hybrid", "d": "2026-02-26", "pr": 43358, "by": "Yanhong Li", "t": "Add OLMo Hybrid model"}, {"m": "pp_doclayout_v2", "d": "2026-02-27", "pr": 43018, "by": "zhang-prog", "t": "[Model] Add PP-DocLayoutV2 Model Support"}, {"m": "timesfm2_5", "d": "2026-02-27", "pr": 41763, "by": "Kashif Rasul", "t": "Timesfm 2.5"}, {"m": "vibevoice_asr", "d": "2026-03-02", "pr": 43625, "by": "Eric Bezzam", "t": "Add VibeVoice ASR"}, {"m": "nemotron_h", "d": "2026-03-03", "pr": 44390, "by": "Li Ding", "t": "add support for nemotron_3"}, {"m": "eurobert", "d": "2026-03-04", "pr": 39455, "by": "Arthur", "t": "Add eurobert"}, {"m": "chmv2", "d": "2026-03-11", "pr": 44595, "by": "Yoni Gozlan", "t": "Add CHMv2"}, {"m": "pp_ocrv5_server_det", "d": "2026-03-13", "pr": 43274, "by": "XingweiDeng", "t": "[Model] Add PP-OCRV5_server_det Model Support"}, {"m": "pp_lcnet", "d": "2026-03-13", "pr": 43247, "by": "XingweiDeng", "t": "[Model] Add PP-OCRV5_mobile_det Model Support"}, {"m": "pp_lcnet_v3", "d": "2026-03-13", "pr": 43247, "by": "XingweiDeng", "t": "[Model] Add PP-OCRV5_mobile_det Model Support"}, {"m": "pp_ocrv5_mobile_det", "d": "2026-03-13", "pr": 43247, "by": "XingweiDeng", "t": "[Model] Add PP-OCRV5_mobile_det Model Support"}, {"m": "pi0", "d": "2026-03-16", "pr": 44160, "by": "Pablo Montalvo", "t": "Add model lerobot PI0 to transformers"}, {"m": "mistral4", "d": "2026-03-16", "pr": 44760, "by": "Julien Denize", "t": "Add Mistral 4"}, {"m": "jina_embeddings_v3", "d": "2026-03-19", "pr": 44251, "by": "Sai-Suraj-27", "t": "Add `Jina-Embeddings-V3` Model"}, {"m": "pp_ocrv5_mobile_rec", "d": "2026-03-19", "pr": 44808, "by": "zhang-prog", "t": "[Model] Add PP-OCRv5_server_rec and PP-OCRv5_mobile_rec models Support"}, {"m": "pp_ocrv5_server_rec", "d": "2026-03-19", "pr": 44808, "by": "zhang-prog", "t": "[Model] Add PP-OCRv5_server_rec and PP-OCRv5_mobile_rec models Support"}, {"m": "pp_chart2table", "d": "2026-03-20", "pr": 43767, "by": "XingweiDeng", "t": "[Model] Add PP-Chart2Table Model Support"}, {"m": "slanext", "d": "2026-03-21", "pr": 43707, "by": "Liu Jiaxuan", "t": "[Model] Add SLANeXt Model Support"}, {"m": "uvdoc", "d": "2026-03-21", "pr": 43385, "by": "XingweiDeng", "t": "[Model] Add UVDoc Model Support"}, {"m": "videomt", "d": "2026-03-25", "pr": 44285, "by": "NielsRogge", "t": "Add VidEoMT"}, {"m": "cohere_asr", "d": "2026-03-26", "pr": 45023, "by": "eustlb", "t": "Add cohere asr"}, {"m": "musicflamingo", "d": "2026-03-30", "pr": 43538, "by": "Lasha Koroshinadze", "t": "Add Music Flamingo"}, {"m": "nomic_bert", "d": "2026-04-02", "pr": 43067, "by": "Sonny Cooper", "t": "Internalise the NomicBERT model"}, {"m": "gemma4", "d": "2026-04-02", "pr": 45192, "by": "Ryan Mullins", "t": "casually dropping the most capable open weights on the planet"}, {"m": "sam3_lite_text", "d": "2026-04-13", "pr": 44320, "by": "NielsRogge", "t": "Add SAM3-LiteText"}, {"m": "qianfan_ocr", "d": "2026-04-17", "pr": 45280, "by": "bairongz", "t": "add Qianfan-OCR model definition"}, {"m": "slanet", "d": "2026-04-22", "pr": 45532, "by": "zhang-prog", "t": "[Model] Add SLANet Model Support"}, {"m": "hy_v3", "d": "2026-04-22", "pr": 45574, "by": "Anton Vlasjuk", "t": "Fix typos"}, {"m": "openai_privacy_filter", "d": "2026-04-22", "pr": 45580, "by": "Anton Vlasjuk", "t": "[`Privacy Filter`] Add model"}, {"m": "deimv2", "d": "2026-04-27", "pr": 44339, "by": "Harshal Janjani", "t": "model: Add DEIMv2 to Transformers"}, {"m": "laguna", "d": "2026-04-28", "pr": 45673, "by": "Joe Rowell", "t": "Laguna XS.2 implementation"}, {"m": "minicpmv4_6", "d": "2026-04-28", "pr": 45689, "by": "Anton Vlasjuk", "t": "fixing more typos"}, {"m": "granite_speech_plus", "d": "2026-04-29", "pr": 45695, "by": "Zvi Kons", "t": "Support for a new Granite-Speech-Plus model"}, {"m": "pp_formulanet", "d": "2026-04-30", "pr": 45626, "by": "zhang-prog", "t": "[Model] Add PP-FormulaNet Model Support"}, {"m": "deepseek_v4", "d": "2026-05-02", "pr": 45643, "by": "Arthur", "t": "Add DeepSeek V4"}, {"m": "exaone4_5", "d": "2026-05-04", "pr": 45471, "by": "Junwon Hwang", "t": "Add EXAONE 4.5 implementations"}, {"m": "gemma4_assistant", "d": "2026-05-05", "pr": 45788, "by": "Sindhu Raghuram Panyam", "t": "First model"}, {"m": "granite4_vision", "d": "2026-05-05", "pr": 45597, "by": "artem-spector", "t": "Add Granite 4.1 Vision (granite4_vision)"}, {"m": "rf_detr", "d": "2026-05-07", "pr": 36895, "by": "StevenBucaille", "t": "Add RF-DETR"}, {"m": "hyperclovax", "d": "2026-05-08", "pr": 44956, "by": "bigshanedogg", "t": "Add HyperCLOVAX SEED Think 14B"}, {"m": "hrm_text", "d": "2026-05-18", "pr": 46025, "by": "yifei wu", "t": "Add hrm text"}, {"m": "pp_lcnet_v4", "d": "2026-05-19", "pr": 45838, "by": "zhang-prog", "t": "[Model] Add PP-OCRv6 Models Support"}, {"m": "pp_ocrv6_medium_det", "d": "2026-05-19", "pr": 45838, "by": "zhang-prog", "t": "[Model] Add PP-OCRv6 Models Support"}, {"m": "pp_ocrv6_small_det", "d": "2026-05-19", "pr": 45838, "by": "zhang-prog", "t": "[Model] Add PP-OCRv6 Models Support"}, {"m": "pp_ocrv6_small_rec", "d": "2026-05-19", "pr": 45838, "by": "zhang-prog", "t": "[Model] Add PP-OCRv6 Models Support"}, {"m": "pp_ocrv6_tiny_rec", "d": "2026-05-19", "pr": 45838, "by": "zhang-prog", "t": "[Model] Add PP-OCRv6 Models Support"}, {"m": "cohere2_moe", "d": "2026-05-20", "pr": 46115, "by": "Cyril Vallez", "t": "Add new cohere2_moe model"}, {"m": "glmga", "d": "2026-05-26", "pr": 46184, "by": "Yuxuan Zhang", "t": "[GLM-4.6V] Update with GLM-GA Processor"}, {"m": "mellum", "d": "2026-05-28", "pr": 46112, "by": "Madeesh Kannan", "t": "feat: Add support for JetBrains' `Mellum` v2 code generation model"}, {"m": "deepseek_ocr2", "d": "2026-06-02", "pr": 45075, "by": "Eon Kim", "t": "Add Deepseek-OCR-2 model"}, {"m": "sapiens2", "d": "2026-06-03", "pr": 45919, "by": "guarin", "t": "Add Sapiens2 Model"}, {"m": "gemma4_unified", "d": "2026-06-03", "pr": 46385, "by": "Douglas Reid", "t": "who needs encoders?"}, {"m": "gemma4_unified_assistant", "d": "2026-06-03", "pr": 46385, "by": "Douglas Reid", "t": "who needs encoders?"}, {"m": "cosmos3_omni", "d": "2026-06-05", "pr": 46146, "by": "MaciejBalaNV", "t": "Added cosmos3 model"}, {"m": "diffusion_gemma", "d": "2026-06-10", "pr": 46540, "by": "Joao Gante", "t": "GPU go brr"}, {"m": "deepseek_v32", "d": "2026-06-11", "pr": 41251, "by": "Arthur", "t": "Add deepseek 3.2 exp"}, {"m": "minimax_m3_vl", "d": "2026-06-12", "pr": 46600, "by": "Arthur", "t": "Add minimax m3vl"}, {"m": "nemotron_asr_streaming", "d": "2026-06-19", "pr": 46332, "by": "eustlb", "t": "Add Nemotron ASR Streaming"}, {"m": "videoprism", "d": "2026-06-20", "pr": 39895, "by": "MAHIR DAIYAN", "t": "Add Videoprism"}, {"m": "minicpm3", "d": "2026-06-23", "pr": 41116, "by": "Minho Ryu", "t": "Add MiniCPM3"}, {"m": "xcodec2", "d": "2026-06-25", "pr": 44178, "by": "Eric Bezzam", "t": "Add xcodec2 model"}, {"m": "qwen3_asr", "d": "2026-06-26", "pr": 43838, "by": "Muhammed", "t": "Qwen3 ASR and Forced Aligner"}, {"m": "nemotron3_5_asr", "d": "2026-06-27", "pr": 46565, "by": "eustlb", "t": "Add Nemotron 3.5 ASR Streaming"}, {"m": "radio", "d": "2026-07-01", "pr": 46425, "by": "meatybobby", "t": "Add support for RADIO models"}, {"m": "zaya", "d": "2026-07-01", "pr": 45862, "by": "JJJYmmm", "t": "[new model] Add Zyphra/ZAYA1-8B"}, {"m": "mimo_v2_flash", "d": "2026-07-01", "pr": 45144, "by": "casinca", "t": "Add Xiaomi MiMo-V2"}, {"m": "hunyuan_vl", "d": "2026-07-03", "pr": 46417, "by": "Mi Jiazhi", "t": "Add HunYuan VL model"}, {"m": "kimi_k25", "d": "2026-07-03", "pr": 45630, "by": "Raushan Turganbay", "t": "Add new model: Kimi2-6"}, {"m": "tipsv2", "d": "2026-07-07", "pr": 46347, "by": "Zixuan Zhu", "t": "Add TIPSv2"}, {"m": "tipsv2_dpt", "d": "2026-07-07", "pr": 46347, "by": "Zixuan Zhu", "t": "Add TIPSv2"}, {"m": "inkling", "d": "2026-07-15", "pr": 47347, "by": "Pablo Montalvo", "t": "Add Inkling model"}, {"m": "cosmos3_edge", "d": "2026-07-16", "pr": 47181, "by": "Atharva Joshi", "t": "Add Cosmos3 Edge model support"}, {"m": "axk1", "d": "2026-07-24", "pr": 46867, "by": "Minsang Kim", "t": "add_axk1"}, {"m": "axk2", "d": "2026-07-25", "pr": 47528, "by": "Anton Vlasjuk", "t": "Add AXK2 from SKT"}, {"m": "granite_swa", "d": "2026-07-29", "pr": 47179, "by": "Davis Wertheimer", "t": "Add Granite-swa and Granitemoe-swa model support"}, {"m": "granitemoe_swa", "d": "2026-07-29", "pr": 47179, "by": "Davis Wertheimer", "t": "Add Granite-swa and Granitemoe-swa model support"}, {"m": "muse_glimmer", "d": "2026-08-10", "pr": 47867, "by": "Cyril Vallez", "t": "Add MuseGlimmer"}, {"m": "muse_glimmer_assistant", "d": "2026-08-10", "pr": 47867, "by": "Cyril Vallez", "t": "Add MuseGlimmer"}, {"m": "cohere_compass", "d": "2026-08-10", "pr": 47878, "by": "calpt", "t": "Add CohereCompass modeling"}, {"m": "step3p7", "d": "2026-08-16", "pr": 46658, "by": "Ita Zaporozhets", "t": "[new model] step 3.7"}, {"m": "esmc", "d": "2026-08-20", "pr": 46419, "by": "Matt", "t": "Port ESMC and ESMFold2 to Transformers"}, {"m": "esmfold2", "d": "2026-08-20", "pr": 46419, "by": "Matt", "t": "Port ESMC and ESMFold2 to Transformers"}, {"m": "granite_speech5", "d": "2026-08-25", "pr": 48288, "by": "eustlb", "t": "gs"}, {"m": "qwen4_exp", "d": "2026-08-26", "pr": 48337, "by": "Cyril Vallez", "t": "Add Qwen4Exp model"}, {"m": "glm5_next", "d": "2026-08-26", "pr": 48342, "by": "Shijin Zhang", "t": "[Glm 5.3 Flash] GLM 5.3 Flash Support"}, {"m": "vibevoice", "d": "2026-08-27", "pr": 40546, "by": "Zhiliang Peng", "t": "Implement VibeVoice"}]; +const MODEL_ROW = {"n": 267, "span": ["2024-06-22", "2026-08-27"], "owners": [["eustlb", 14], ["Arthur", 12], ["zhang-prog", 12], ["Yoni Gozlan", 10]], "authors": 126}; const TRACKS = [ { @@ -15,30 +18,54 @@ const TRACKS = [ "tag": "3 frameworks → 1", "s": "done", "prs": [ - [ - 38760, - "2025-06-11", - "Prepare for TF+Jax deprecation", - "MERGED" - ], - [ - 40654, - "2025-09-03", - "Remove TF/Flax examples", - "MERGED" - ], - [ - 40760, - "2025-09-18", - "🚨🚨🚨 Fully remove Tensorflow and Jax support library-wide", - "MERGED" - ], - [ - 41058, - "2025-09-23", - "Remove mention of TensorFlow/Flax/JAX from English documentation", - "MERGED" - ] + { + "n": 38760, + "d": "2025-06-11", + "t": "Prepare for TF+Jax deprecation", + "st": "MERGED", + "by": "Rocketknight1", + "a": 4, + "r": 37, + "f": 2, + "c": 2, + "b": "This PR removes a lot of TF + Jax testing in repo-consistency and .circleci. In a follow-up PR, I'll delete a lot of the test files, but this needs to be in a separate PR because repo-consistency checks run against the target branch, not the PR branch, and so they will fail if we remove other tests " + }, + { + "n": 40654, + "d": "2025-09-03", + "t": "Remove TF/Flax examples", + "st": "MERGED", + "by": "Rocketknight1", + "a": 1, + "r": 21961, + "f": 74, + "c": 3, + "b": "As the title says, the frameworks have been deprecated, so we should probably drop the examples as well! One thing to note: check_copies.py had a check_full_copies function that checked when files should be exact copies of each other. This was only used for TF/Flax examples. I've removed the FULL_CO" + }, + { + "n": 40760, + "d": "2025-09-18", + "t": "🚨🚨🚨 Fully remove Tensorflow and Jax support library-wide", + "st": "MERGED", + "by": "Cyrilvallez", + "a": 3792, + "r": 181269, + "f": 854, + "c": 35, + "b": "Apart from obvious tf/jax support, I believe the following should be the only potential breaking changes to torch-only code: pipelines do not take framework argument anymore onnx config methods do not take framework argument anymore It may break current torch code if users do framework=\"pt\" explicit" + }, + { + "n": 41058, + "d": "2025-09-23", + "t": "Remove mention of TensorFlow/Flax/JAX from English documentation", + "st": "MERGED", + "by": "cyyever", + "a": 3, + "r": 4, + "f": 2, + "c": 1, + "b": "There is a remaining one. Before submitting This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case). Did you read the contributor guideline, Pull Request section? Was this discussed/approved via a Github issue or the forum? Please add a link to it if that's th" + } ], "note": null, "links": [], @@ -46,19 +73,46 @@ const TRACKS = [ "span": [ "2025-06-11", "2025-09-23" - ] + ], + "roll": { + "owners": [ + [ + "Rocketknight1", + 2 + ], + [ + "Cyrilvallez", + 1 + ], + [ + "cyyever", + 1 + ] + ], + "n": 4, + "add": 3800, + "del": 203271, + "files": 932, + "commits": 41 + } }, { "t": "Slow tokenizers removed", "tag": "fast tokenizers only", "s": "done", "prs": [ - [ - 40936, - "2025-11-27", - "rm slow tokenizers", - "MERGED" - ] + { + "n": 40936, + "d": "2025-11-27", + "t": "rm slow tokenizers", + "st": "MERGED", + "by": "itazap", + "a": 16908, + "r": 55397, + "f": 443, + "c": 277, + "b": "Tokenization Just as we moved towards a single backend library for model definition, we want Tokenizer to be a lot more intuitive. With v5, you can now initialize an empty LlamaTokenizer and train it directly on your new task! Defining a new tokenizer object should be as simple as this: from transfo" + } ], "note": null, "links": [], @@ -66,43 +120,86 @@ const TRACKS = [ "span": [ "2025-10-10", "2025-11-27" - ] + ], + "roll": { + "owners": [ + [ + "itazap", + 1 + ] + ], + "n": 1, + "add": 16908, + "del": 55397, + "files": 443, + "commits": 277 + } }, { "t": "v5.0.0 released", "tag": "v5.0.0 · 2026-01-26", "s": "done", "prs": [ - [ - 40997, - "2025-09-19", - "🚨 [v5] remove deprecated entry point", - "MERGED" - ], - [ - 40998, - "2025-09-19", - "🚨 [v5] remove generate output retrocompatibility aliases", - "MERGED" - ], - [ - 41017, - "2025-09-24", - "🚨 [V5] Remove deprecated training arguments ", - "MERGED" - ], - [ - 41076, - "2025-09-30", - ":rotating_light: [`v5`] Remove headmasking", - "MERGED" - ], - [ - 41123, - "2025-09-30", - "[v5] Remove deprecated prediction loop", - "MERGED" - ] + { + "n": 40997, + "d": "2025-09-19", + "t": "🚨 [v5] remove deprecated entry point", + "st": "MERGED", + "by": "gante", + "a": 99, + "r": 111, + "f": 30, + "c": 3, + "b": "Removes the transformers-cli entrypoint, which is deprecated in favor of transformers" + }, + { + "n": 40998, + "d": "2025-09-19", + "t": "🚨 [v5] remove generate output retrocompatibility aliases", + "st": "MERGED", + "by": "gante", + "a": 0, + "r": 66, + "f": 3, + "c": 2, + "b": "See title :)" + }, + { + "n": 41017, + "d": "2025-09-24", + "t": "🚨 [V5] Remove deprecated training arguments ", + "st": "MERGED", + "by": "cyyever", + "a": 12, + "r": 234, + "f": 5, + "c": 4, + "b": "Remove deprecated training arguments from TrainingArguments" + }, + { + "n": 41076, + "d": "2025-09-30", + "t": ":rotating_light: [`v5`] Remove headmasking", + "st": "MERGED", + "by": "vasqu", + "a": 1154, + "r": 8688, + "f": 497, + "c": 24, + "b": "As per title, time to deprecate this This includes anything in the docs, tests, and modeling xxx head mask xxx_headmasking xxx_head_mask" + }, + { + "n": 41123, + "d": "2025-09-30", + "t": "[v5] Remove deprecated prediction loop", + "st": "MERGED", + "by": "SunMarc", + "a": 6, + "r": 530, + "f": 11, + "c": 5, + "b": "This PR removes the legacy prediction loop that was deprecated 4 years ago. I think that it should be fine to deprecate it even without any warning for v5 which it would have been better to have. I checked files on github and no one really seemed to use the legacy one. Also SequentialDistributedSamp" + } ], "note": "Board placed v5 in Q2 ’25. v5.0.0rc0 tagged 2025-12-01, v5.0.0 released 2026-01-26 — nine months later.", "links": [], @@ -110,7 +207,32 @@ const TRACKS = [ "span": [ "2025-09-19", "2026-01-26" - ] + ], + "roll": { + "owners": [ + [ + "gante", + 2 + ], + [ + "cyyever", + 1 + ], + [ + "vasqu", + 1 + ], + [ + "SunMarc", + 1 + ] + ], + "n": 5, + "add": 1271, + "del": 9629, + "files": 546, + "commits": 38 + } } ] }, @@ -124,24 +246,42 @@ const TRACKS = [ "tag": "one attention API for every model", "s": "done", "prs": [ - [ - 35235, - "2024-12-18", - "🚨All attention refactor🚨", - "MERGED" - ], - [ - 35342, - "2024-12-20", - "Cleaner attention interfaces", - "MERGED" - ], - [ - 38108, - "2025-05-22", - "🔴🔴🔴 [`Attention`] Refactor Attention Interface for Bart-based Models", - "MERGED" - ] + { + "n": 35235, + "d": "2024-12-18", + "t": "🚨All attention refactor🚨", + "st": "MERGED", + "by": "ArthurZucker", + "a": 5635, + "r": 9778, + "f": 107, + "c": 99, + "b": "Todo in this PR: Cohere Chameleon DBRX Gemma Gemma2 GLM (modular donc rien à faire je crois) gpt_neoX et GPT2 Granite Jamba JetMoe Mimi Mistral Mixtral Mllama Moshi Nemotron OPT Phi Ph3 PhiMoe Qwen2 qwen2Moe qwen2VL SableML StartCoder2 -> Modular normalement oK Idefics1,2,3 Olmo Olmo2 Siglip Whisper" + }, + { + "n": 35342, + "d": "2024-12-20", + "t": "Cleaner attention interfaces", + "st": "MERGED", + "by": "Cyrilvallez", + "a": 30, + "r": 12, + "f": 4, + "c": 5, + "b": "Improve the attention interfaces with comments/dtype fix" + }, + { + "n": 38108, + "d": "2025-05-22", + "t": "🔴🔴🔴 [`Attention`] Refactor Attention Interface for Bart-based Models", + "st": "MERGED", + "by": "vasqu", + "a": 8035, + "r": 5932, + "f": 75, + "c": 71, + "b": "This PR is gonna tackle two things in general: Flex Attention for all base attention types (encoder, decoder, encoder-decoder cross) New Attention Interface for a bunch of models (mostly based on Bart's implementation) As a bonus, some models have already been refactored into modular if seen fit, e." + } ], "note": null, "links": [], @@ -149,25 +289,58 @@ const TRACKS = [ "span": [ "2024-12-16", "2025-05-22" - ] + ], + "roll": { + "owners": [ + [ + "ArthurZucker", + 1 + ], + [ + "Cyrilvallez", + 1 + ], + [ + "vasqu", + 1 + ] + ], + "n": 3, + "add": 13700, + "del": 15722, + "files": 186, + "commits": 175 + } }, { "t": "Flex Attention", "tag": "torch flex backend", "s": "done", "prs": [ - [ - 34896, - "2024-12-04", - "[`GPTNeoX`] Flex Attention + Refactor", - "MERGED" - ], - [ - 37381, - "2025-04-14", - "Fix mask handling for flex attention in llama/gemma2/mistral/qwen2", - "MERGED" - ] + { + "n": 34896, + "d": "2024-12-04", + "t": "[`GPTNeoX`] Flex Attention + Refactor", + "st": "MERGED", + "by": "vasqu", + "a": 371, + "r": 249, + "f": 6, + "c": 20, + "b": "Adds flex attention and the refactor according to #34809 However, I discovered several issues in the current version of gemma2 (#34282): It seems like that flex attention needs a transpose afterwards like sdpa Loading flex attn with from pretrained didn't work and hence, current tests use another at" + }, + { + "n": 37381, + "d": "2025-04-14", + "t": "Fix mask handling for flex attention in llama/gemma2/mistral/qwen2", + "st": "MERGED", + "by": "flukeskywalker", + "a": 423, + "r": 177, + "f": 59, + "c": 4, + "b": "Issue In gemma2 and mistral modelling code, where the _update_causal_mask() methods were not handling the case of the atn_implementation being the recently added \"flex_attention\". Main Fix For Mistral and Gemma2: Added logic for converting attention mask passed as as a tensor to a BlockMask required" + } ], "note": null, "links": [], @@ -175,43 +348,90 @@ const TRACKS = [ "span": [ "2024-11-23", "2025-04-14" - ] + ], + "roll": { + "owners": [ + [ + "vasqu", + 1 + ], + [ + "flukeskywalker", + 1 + ] + ], + "n": 2, + "add": 794, + "del": 426, + "files": 65, + "commits": 24 + } }, { "t": "torch.compile coverage", "tag": "compile across the model zoo", "s": "done", "prs": [ - [ - 31247, - "2024-07-18", - "Add torch.compile Support For Mamba", - "MERGED" - ], - [ - 32227, - "2024-07-31", - ">3-5x faster torch.compile forward compilation for autoregressive decoder models", - "MERGED" - ], - [ - 35187, - "2024-12-23", - "make LlamaModel._update_causal_mask torch compilable", - "MERGED" - ], - [ - 35658, - "2025-01-13", - "[`Compile`] Only test compiling model forward pass", - "MERGED" - ], - [ - 34544, - "2025-01-28", - "Test: generate with `torch.compile(model.forward)` as a fast test", - "MERGED" - ] + { + "n": 31247, + "d": "2024-07-18", + "t": "Add torch.compile Support For Mamba", + "st": "MERGED", + "by": "zhenglongjiepheonix", + "a": 225, + "r": 85, + "f": 4, + "c": 41, + "b": "torch.compile support for mamba! Closes #31246" + }, + { + "n": 32227, + "d": "2024-07-31", + "t": ">3-5x faster torch.compile forward compilation for autoregressive decoder models", + "st": "MERGED", + "by": "fxmarty", + "a": 1602, + "r": 424, + "f": 22, + "c": 19, + "b": "As detailed in #29114 and #29165, the current implementation of GenerationMixin.generate passes 2D integer tensors of variable length as attention_mask to the models forward, that are transformed into 4D tensors of static shape in the models forward for decoder models. This is an issue when using to" + }, + { + "n": 35187, + "d": "2024-12-23", + "t": "make LlamaModel._update_causal_mask torch compilable", + "st": "MERGED", + "by": "winglian", + "a": 33, + "r": 33, + "f": 33, + "c": 3, + "b": "This doesn't really have a performance improvement, but at least keeps torch.compile from having a graph break here. Before submitting This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case). Did you read the contributor guideline, Pull Request section? Was th" + }, + { + "n": 35658, + "d": "2025-01-13", + "t": "[`Compile`] Only test compiling model forward pass", + "st": "MERGED", + "by": "ArthurZucker", + "a": 9, + "r": 34, + "f": 7, + "c": 2, + "b": "Instead of testing the full generate compile." + }, + { + "n": 34544, + "d": "2025-01-28", + "t": "Test: generate with `torch.compile(model.forward)` as a fast test", + "st": "MERGED", + "by": "gante", + "a": 105, + "r": 53, + "f": 25, + "c": 13, + "b": "Follow-up to #34464 This PR: Converts test_generate_compile_model_forward to a fast test. This means we will check generate with torch.compile(model.forward) at each commit on ALL models that support StaticCache 💛 Fixes failing cases of test_generate_compile_model_forward whenever possible Tags mode" + } ], "note": null, "links": [], @@ -219,43 +439,98 @@ const TRACKS = [ "span": [ "2024-06-04", "2025-01-28" - ] + ], + "roll": { + "owners": [ + [ + "zhenglongjiepheonix", + 1 + ], + [ + "fxmarty", + 1 + ], + [ + "winglian", + 1 + ], + [ + "ArthurZucker", + 1 + ] + ], + "n": 5, + "add": 1974, + "del": 629, + "files": 91, + "commits": 78 + } }, { "t": "Kernels from the Hub (`kernelize`)", "tag": "community kernels, no rebuild", "s": "done", "prs": [ - [ - 36853, - "2025-03-21", - "Use `deformable_detr` kernel from the Hub", - "MERGED" - ], - [ - 38205, - "2025-06-24", - "Add kernelize to transformers", - "MERGED" - ], - [ - 41657, - "2025-10-16", - "Erroring when KernelConfig is passed without use_kernels = True", - "MERGED" - ], - [ - 41990, - "2025-11-12", - "feat(kernels): add opt-out flag to disable kernels hub usage through the lib", - "MERGED" - ], - [ - 47863, - "2026-08-17", - "Bump default flash-attn2 hub kernel version to v3", - "MERGED" - ] + { + "n": 36853, + "d": "2025-03-21", + "t": "Use `deformable_detr` kernel from the Hub", + "st": "MERGED", + "by": "danieldk", + "a": 405, + "r": 3834, + "f": 21, + "c": 3, + "b": "Remove the deformable_detr kernel from kernels/ and use the pre-built kernel from the Hub instead. Before submitting This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case). Did you read the contributor guideline, Pull Request section? Was this discussed/appro" + }, + { + "n": 38205, + "d": "2025-06-24", + "t": "Add kernelize to transformers", + "st": "MERGED", + "by": "MekkCyber", + "a": 13, + "r": 43, + "f": 4, + "c": 11, + "b": "Instead of dynamically switching the forward methods using a decorator, we are exploring a new approach that performs this replacement statically within modeling_utils.py. This allows us to modify the forward methods at load time, which makes the kernels compile compatible. Also there is no need to " + }, + { + "n": 41657, + "d": "2025-10-16", + "t": "Erroring when KernelConfig is passed without use_kernels = True", + "st": "MERGED", + "by": "MekkCyber", + "a": 6, + "r": 0, + "f": 1, + "c": 3, + "b": "as per title" + }, + { + "n": 41990, + "d": "2025-11-12", + "t": "feat(kernels): add opt-out flag to disable kernels hub usage through the lib", + "st": "MERGED", + "by": "mfuntowicz", + "a": 138, + "r": 2, + "f": 2, + "c": 7, + "b": "This PR attempts to bring an opt-out mecanism to disable hub kernels (aka kernels) globally through transformers. The default value for the USE_HUB_KERNELS environment variable defaults to YES to maintain backward compatiblity with the current machinery. In more details, it works by proxying the ker" + }, + { + "n": 47863, + "d": "2026-08-17", + "t": "Bump default flash-attn2 hub kernel version to v3", + "st": "MERGED", + "by": "jiqing-feng", + "a": 19, + "r": 3, + "f": 1, + "c": 6, + "b": "What load_and_register_attn_kernel hardcodes version=1 as the default major version for every attention kernel loaded from the Hub. Pin kernels-community/flash-attn2 to v3 through a small per-repo table instead. Why The v1 branch of kernels-community/flash-attn2 only ships builds up to Torch 2.12, s" + } ], "note": null, "links": [], @@ -263,61 +538,134 @@ const TRACKS = [ "span": [ "2025-03-21", "2026-08-17" - ] + ], + "roll": { + "owners": [ + [ + "MekkCyber", + 2 + ], + [ + "danieldk", + 1 + ], + [ + "mfuntowicz", + 1 + ], + [ + "jiqing-feng", + 1 + ] + ], + "n": 5, + "add": 581, + "del": 3882, + "files": 29, + "commits": 30 + } }, { "t": "MoE / expert kernels", "tag": "grouped_mm + batched experts", "s": "done", "prs": [ - [ - 37376, - "2025-04-11", - "Add moe kernels", - "MERGED" - ], - [ - 42697, - "2026-01-05", - "batched and grouped experts implementations", - "MERGED" - ], - [ - 43227, - "2026-01-22", - "GptOss experts implementation", - "MERGED" - ], - [ - 43977, - "2026-02-16", - "stable grouped_mm API", - "MERGED" - ], - [ - 44043, - "2026-02-18", - "`grouped_mm` fallback", - "MERGED" - ], - [ - 44152, - "2026-02-20", - "AutoGrad support for grouped_mm fallback", - "MERGED" - ], - [ - 45621, - "2026-05-04", - "Better Grouped GEMM + EP", - "MERGED" - ], - [ - 45634, - "2026-06-02", - "DeepGEMM BF16 + mixed FP8/FP4 + MegaMoE + refactor", - "MERGED" - ] + { + "n": 37376, + "d": "2025-04-11", + "t": "Add moe kernels", + "st": "MERGED", + "by": "ArthurZucker", + "a": 9, + "r": 0, + "f": 2, + "c": 10, + "b": "Add MoE kernels" + }, + { + "n": 42697, + "d": "2026-01-05", + "t": "batched and grouped experts implementations", + "st": "MERGED", + "by": "IlyasMoutawwakil", + "a": 1060, + "r": 374, + "f": 65, + "c": 74, + "b": "I have started experimenting with pure pytorch MoE implementations following the HF exporters PR while trying to find a traceable/exportable variant for onnx/openvino. In this PR I copy the attn_implementation API into a similar experts_implementation API, and added two new implementations: batched_" + }, + { + "n": 43227, + "d": "2026-01-22", + "t": "GptOss experts implementation", + "st": "MERGED", + "by": "IlyasMoutawwakil", + "a": 359, + "r": 223, + "f": 8, + "c": 35, + "b": "[wip] Fixes #43193 @vasqu @mattteochen Before submitting This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case). Did you read the contributor guideline, Pull Request section? Was this discussed/approved via a Github issue or the forum? Please add a link to it" + }, + { + "n": 43977, + "d": "2026-02-16", + "t": "stable grouped_mm API", + "st": "MERGED", + "by": "IlyasMoutawwakil", + "a": 48, + "r": 23, + "f": 5, + "c": 9, + "b": "Fixes # (issue) Before submitting This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case). Did you read the contributor guideline, Pull Request section? Was this discussed/approved via a Github issue or the forum? Please add a link to it if that's the case. Di" + }, + { + "n": 44043, + "d": "2026-02-18", + "t": "`grouped_mm` fallback", + "st": "MERGED", + "by": "IlyasMoutawwakil", + "a": 170, + "r": 162, + "f": 31, + "c": 8, + "b": "With this PR, grouped_mm impl will always be default when the model uses experts impl (even in old torch versions) and will no longer fail if you are torch compiling an fp16/fp32 model, the impl will succeed in all cases, except cuda-graphs (common failure case in all grouped_mm impls). Before submi" + }, + { + "n": 44152, + "d": "2026-02-20", + "t": "AutoGrad support for grouped_mm fallback", + "st": "MERGED", + "by": "IlyasMoutawwakil", + "a": 79, + "r": 45, + "f": 3, + "c": 6, + "b": "Fixes # (issue) Before submitting This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case). Did you read the contributor guideline, Pull Request section? Was this discussed/approved via a Github issue or the forum? Please add a link to it if that's the case. Di" + }, + { + "n": 45621, + "d": "2026-05-04", + "t": "Better Grouped GEMM + EP", + "st": "MERGED", + "by": "IlyasMoutawwakil", + "a": 394, + "r": 219, + "f": 5, + "c": 36, + "b": "The idea is that we shouldn't be clamping experts ids at all, clamping them makes their tokens get prjected as if they were routed to the last expert, instead, we should let the sentinels be: moved to the queue by the sorting (num_experts/sentinel is the biggest value) dropped by the histogram (beca" + }, + { + "n": 45634, + "d": "2026-06-02", + "t": "DeepGEMM BF16 + mixed FP8/FP4 + MegaMoE + refactor", + "st": "MERGED", + "by": "IlyasMoutawwakil", + "a": 1627, + "r": 486, + "f": 16, + "c": 164, + "b": "Based on #45621 GroupedLinear co-authored-by @sywangyi Code Agent Policy The Transformers repo is currently being overwhelmed by a large number of PRs and issue comments written by code agents. We are currently bottlenecked by our ability to review and respond to them. As a result, we ask that new u" + } ], "note": null, "links": [], @@ -325,49 +673,102 @@ const TRACKS = [ "span": [ "2025-04-08", "2026-06-02" - ] + ], + "roll": { + "owners": [ + [ + "IlyasMoutawwakil", + 7 + ], + [ + "ArthurZucker", + 1 + ] + ], + "n": 8, + "add": 3746, + "del": 1532, + "files": 135, + "commits": 342 + } }, { "t": "Continuous batching", "tag": "serving-grade batching in-library", "s": "done", "prs": [ - [ - 39457, - "2025-07-22", - "[Paged-Attention] Handle continuous batching for repetition penalty", - "MERGED" - ], - [ - 40426, - "2025-08-26", - "Continuous batching refactor", - "MERGED" - ], - [ - 41370, - "2025-10-06", - "[`CB`] Refactors the way we access paged", - "MERGED" - ], - [ - 41655, - "2025-10-16", - "Switch to CB if cache_implementation == paged", - "MERGED" - ], - [ - 41916, - "2025-11-07", - "feat(ci): add continuous batching to benchmarks", - "MERGED" - ], - [ - 43521, - "2026-01-28", - "[CB] Minor perf improvements and ty compatibility", - "MERGED" - ] + { + "n": 39457, + "d": "2025-07-22", + "t": "[Paged-Attention] Handle continuous batching for repetition penalty", + "st": "MERGED", + "by": "kashif", + "a": 80, + "r": 5, + "f": 3, + "c": 9, + "b": "We handle the repetition penalty for continuous batching Fixes # (issue) Before submitting This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case). Did you read the contributor guideline, Pull Request section? Was this discussed/approved via a Github issue or " + }, + { + "n": 40426, + "d": "2025-08-26", + "t": "Continuous batching refactor", + "st": "MERGED", + "by": "remi-or", + "a": 2174, + "r": 1547, + "f": 10, + "c": 27, + "b": "This PR refactors the files related to continuous batching and changes a few parts of the logic: the way the max_batch_tokens and num_blocks is computed; the static inputs of the CB are now sliced to avoid passing an arbitrary amount of padding in the forward the example file for CB has been reworke" + }, + { + "n": 41370, + "d": "2025-10-06", + "t": "[`CB`] Refactors the way we access paged", + "st": "MERGED", + "by": "ArthurZucker", + "a": 33, + "r": 43, + "f": 8, + "c": 6, + "b": "The user decides on the function, but CB handles itself which \"interface\" wrapper to use. This should make stuff easier" + }, + { + "n": 41655, + "d": "2025-10-16", + "t": "Switch to CB if cache_implementation == paged", + "st": "MERGED", + "by": "remi-or", + "a": 75, + "r": 3, + "f": 2, + "c": 7, + "b": "This PR fixes one of the issues raised in #41501 This PR adds paged as a static cache implementation and switches from generate to generate_batch if it's the chosen implementation. It also raises errors if an argument for generate is passed to generate_batch but is not supported, and warn the user t" + }, + { + "n": 41916, + "d": "2025-11-07", + "t": "feat(ci): add continuous batching to benchmarks", + "st": "MERGED", + "by": "McPatate", + "a": 190, + "r": 107, + "f": 10, + "c": 15, + "b": "" + }, + { + "n": 43521, + "d": "2026-01-28", + "t": "[CB] Minor perf improvements and ty compatibility", + "st": "MERGED", + "by": "remi-or", + "a": 228, + "r": 169, + "f": 8, + "c": 21, + "b": "Summary This PR introduces minor changes to the continuous batching feature: Performance optimize the get_seqlens_k function to avoid looping over all cache managers logits indexing is now done inside the cuda graph, as there was no apparent reason it should be done outside fuse the get_read/write_i" + } ], "note": null, "links": [], @@ -375,43 +776,98 @@ const TRACKS = [ "span": [ "2025-07-16", "2026-01-28" - ] + ], + "roll": { + "owners": [ + [ + "remi-or", + 3 + ], + [ + "kashif", + 1 + ], + [ + "ArthurZucker", + 1 + ], + [ + "McPatate", + 1 + ] + ], + "n": 6, + "add": 2780, + "del": 1874, + "files": 41, + "commits": 85 + } }, { "t": "`transformers serve` + chat CLI", "tag": "OpenAI-compatible endpoint", "s": "done", "prs": [ - [ - 38443, - "2025-06-30", - "Split `transformers chat` and `transformers serve` ", - "MERGED" - ], - [ - 39155, - "2025-07-16", - "Responses API in `transformers serve`", - "MERGED" - ], - [ - 39454, - "2025-07-23", - "Transformers serve VLM", - "MERGED" - ], - [ - 41446, - "2025-10-15", - "Enable non-streaming mode in `transformers serve`", - "MERGED" - ], - [ - 44558, - "2026-04-22", - "Add /v1/completions endpoint (OpenAI legacy completions API) to `transformers serve`", - "MERGED" - ] + { + "n": 38443, + "d": "2025-06-30", + "t": "Split `transformers chat` and `transformers serve` ", + "st": "MERGED", + "by": "LysandreJik", + "a": 915, + "r": 310, + "f": 9, + "c": 25, + "b": "This PR splits the transformers chat frontend from its backend (now moved to transformers serve). We take this opportunity to have transformers serve act as a server with an OpenAI-compatible API, via http, using SSE to stream tokens. You can take it for a spin either by spawning a server: transform" + }, + { + "n": 39155, + "d": "2025-07-16", + "t": "Responses API in `transformers serve`", + "st": "MERGED", + "by": "LysandreJik", + "a": 948, + "r": 391, + "f": 8, + "c": 10, + "b": "This PR offers an initial scaffolding of transformers serve so that it has both /v1/chat/completions and /v1/responses endpoints. So far, it only implements a streaming version of both endpoints. To implement the responses endpoint, I'm basing myself off of the \"Streaming\" tab in the OpenAPI API ref" + }, + { + "n": 39454, + "d": "2025-07-23", + "t": "Transformers serve VLM", + "st": "MERGED", + "by": "LysandreJik", + "a": 268, + "r": 75, + "f": 2, + "c": 8, + "b": "Add VLM support for transformers serve. In case you're interested in playing around with it, Open WebUI is a good tool to work with. Launching transformers serve with VLM support: cd git checkout transformers-serve-vlm transformers serve --enable_cors To install Open WebUI: pi" + }, + { + "n": 41446, + "d": "2025-10-15", + "t": "Enable non-streaming mode in `transformers serve`", + "st": "MERGED", + "by": "LysandreJik", + "a": 211, + "r": 73, + "f": 4, + "c": 3, + "b": "Needs this to be merged first: #41444 Tests and docs need to be added before undraft" + }, + { + "n": 44558, + "d": "2026-04-22", + "t": "Add /v1/completions endpoint (OpenAI legacy completions API) to `transformers serve`", + "st": "MERGED", + "by": "rain-1", + "a": 567, + "r": 1, + "f": 5, + "c": 1, + "b": "Adds support for the legacy text completions endpoint, which accepts a freeform text prompt (no chat template) and returns generated text in choices[].text. Supports both streaming and non-streaming modes, suffix for fill-in-the-middle insertion, and proper finish_reason detection. What does this PR" + } ], "note": null, "links": [], @@ -419,37 +875,78 @@ const TRACKS = [ "span": [ "2025-06-24", "2026-04-22" - ] + ], + "roll": { + "owners": [ + [ + "LysandreJik", + 4 + ], + [ + "rain-1", + 1 + ] + ], + "n": 5, + "add": 2909, + "del": 850, + "files": 28, + "commits": 47 + } }, { "t": "vLLM backend enablers", "tag": "vLLM runs transformers models", "s": "done", "prs": [ - [ - 36934, - "2025-05-02", - "Aligning modling code for GPT2 to work with vLLM (fallback)", - "MERGED" - ], - [ - 37915, - "2025-05-27", - "[transformers x vLLM] standardize processors", - "MERGED" - ], - [ - 39583, - "2025-07-24", - "Update recent processors for vLLM backend", - "MERGED" - ], - [ - 44469, - "2026-03-05", - "[remote code/vllm] Fix incorrect tied weights", - "MERGED" - ] + { + "n": 36934, + "d": "2025-05-02", + "t": "Aligning modling code for GPT2 to work with vLLM (fallback)", + "st": "MERGED", + "by": "ariG23498", + "a": 13, + "r": 0, + "f": 3, + "c": 23, + "b": "This PR changes the modeling code for GPT2 to support working on vLLM using the transformers fallback backend. The changes are as follows: Introduction of kwargs which is used to propagate information about attention_indices in vLLM Adding a base_model_tp_plan, which is currently empty. This is a re" + }, + { + "n": 37915, + "d": "2025-05-27", + "t": "[transformers x vLLM] standardize processors", + "st": "MERGED", + "by": "zucchini-nlp", + "a": 1167, + "r": 79, + "f": 28, + "c": 19, + "b": "Part of #37780. The design was tested on different model types: Oke, I verified that the inference works for all models, unless I forgot about some new ones. Here is the list I tested. A few models (blip2, gotOcr, gemma3) won't be supported in the first release. Gemma3 is already planned after we me" + }, + { + "n": 39583, + "d": "2025-07-24", + "t": "Update recent processors for vLLM backend", + "st": "MERGED", + "by": "zucchini-nlp", + "a": 79, + "r": 9, + "f": 6, + "c": 2, + "b": "As peer title, models needed tiny changes here and there" + }, + { + "n": 44469, + "d": "2026-03-05", + "t": "[remote code/vllm] Fix incorrect tied weights", + "st": "MERGED", + "by": "Cyrilvallez", + "a": 4, + "r": 2, + "f": 1, + "c": 1, + "b": "For remote code that behave correctly with tied weights, we need to keep the same behavior as for the main lib, i.e. not remove them from tied weights (as tied weights are marked as missing to avoid inits!!)" + } ], "note": "The backend itself lives in vllm-project/vllm; these are the transformers-side changes it needs.", "links": [ @@ -462,7 +959,28 @@ const TRACKS = [ "span": [ "2025-03-24", "2026-03-05" - ] + ], + "roll": { + "owners": [ + [ + "zucchini-nlp", + 2 + ], + [ + "ariG23498", + 1 + ], + [ + "Cyrilvallez", + 1 + ] + ], + "n": 4, + "add": 1263, + "del": 90, + "files": 38, + "commits": 45 + } }, { "t": "TGI backend", @@ -477,31 +995,57 @@ const TRACKS = [ ] ], "d": null, - "span": null + "span": null, + "roll": { + "owners": [], + "n": 0, + "add": 0, + "del": 0, + "files": 0, + "commits": 0 + } }, { "t": "Fast image processors by default", "tag": "GPU image preprocessing", "s": "done", "prs": [ - [ - 28847, - "2024-06-11", - "Fast image processor", - "MERGED" - ], - [ - 35184, - "2024-12-23", - "Add compile test for fast image processor", - "MERGED" - ], - [ - 41388, - "2026-01-21", - "🚨Default to fast image processors for all models", - "MERGED" - ] + { + "n": 28847, + "d": "2024-06-11", + "t": "Fast image processor", + "st": "MERGED", + "by": "amyeroberts", + "a": 1643, + "r": 811, + "f": 64, + "c": 40, + "b": "Adds a ViTImageProcessorFast class, which uses a torchvision backend to speed up image processor import requests from PIL import Image from transformers import AutoImageProcessor # Load the image image = Image.open(requests.get(\"http://images.cocodataset.org/val2017/000000039769.jpg\", stream=True).r" + }, + { + "n": 35184, + "d": "2024-12-23", + "t": "Add compile test for fast image processor", + "st": "MERGED", + "by": "yonigozlan", + "a": 63, + "r": 4, + "f": 3, + "c": 4, + "b": "Add a test in ImageProcessingTestMixin, which is only ran on fast image processors. The test compiles a fast image processor, run a processing pass and compares the result with an eager image processor. Who can review?" + }, + { + "n": 41388, + "d": "2026-01-21", + "t": "🚨Default to fast image processors for all models", + "st": "MERGED", + "by": "yonigozlan", + "a": 404, + "r": 259, + "f": 64, + "c": 81, + "b": "Following the trial testing with Qwen_VL image processors, this extends defaulting to fast image processors even for checkpoints saved with a slow one to all models. Also made sure that all processors use AutoImageProcessor to instantiate their image_processor_class. On that point, defining default " + } ], "note": null, "links": [], @@ -509,7 +1053,24 @@ const TRACKS = [ "span": [ "2024-06-10", "2026-01-21" - ] + ], + "roll": { + "owners": [ + [ + "yonigozlan", + 2 + ], + [ + "amyeroberts", + 1 + ] + ], + "n": 3, + "add": 2110, + "del": 1074, + "files": 131, + "commits": 125 + } } ] }, @@ -523,18 +1084,30 @@ const TRACKS = [ "tag": "inheritance for model files", "s": "done", "prs": [ - [ - 33248, - "2024-09-24", - "Modular `transformers`: modularity and inheritance for new model additions", - "MERGED" - ], - [ - 34487, - "2024-11-01", - "Large modular logic refactoring", - "MERGED" - ] + { + "n": 33248, + "d": "2024-09-24", + "t": "Modular `transformers`: modularity and inheritance for new model additions", + "st": "MERGED", + "by": "ArthurZucker", + "a": 6504, + "r": 778, + "f": 41, + "c": 167, + "b": "Modular transformers: supporting inheritance, while keeping our philosophy This PR introduces Modularity for transformers, which has always been prohibited when working with transformers (see blog post for the accompanying design philosophy). The core idea behind this PR is to facilitate model addit" + }, + { + "n": 34487, + "d": "2024-11-01", + "t": "Large modular logic refactoring", + "st": "MERGED", + "by": "Cyrilvallez", + "a": 2668, + "r": 1600, + "f": 19, + "c": 40, + "b": "This PR largely rework the logic we use in the modular converter. It is (hopefully) clearer and maintainable. Instead of going in all directions, adding stuff, then deleting it if not needed, we now do the following: visit all the modular file (record imports/functions/classes/assignments nodes) cre" + } ], "note": null, "links": [], @@ -542,19 +1115,42 @@ const TRACKS = [ "span": [ "2024-09-02", "2024-11-01" - ] + ], + "roll": { + "owners": [ + [ + "ArthurZucker", + 1 + ], + [ + "Cyrilvallez", + 1 + ] + ], + "n": 2, + "add": 9172, + "del": 2378, + "files": 60, + "commits": 207 + } }, { "t": "Modular detector", "tag": "finds what should be modular", "s": "done", "prs": [ - [ - 41289, - "2025-10-03", - "Add modular detector", - "MERGED" - ] + { + "n": 41289, + "d": "2025-10-03", + "t": "Add modular detector", + "st": "MERGED", + "by": "molbap", + "a": 677, + "r": 1, + "f": 3, + "c": 7, + "b": "Adds a new util to parse a new modeling_* file, whether or not it is from transformers, and lists the closest possible matches from existing models in the library using embeddings from Qwen3. Why? Because most models should be able to be defined as modulars from each other. It's not exactly the case" + } ], "note": null, "links": [], @@ -562,19 +1158,38 @@ const TRACKS = [ "span": [ "2025-10-02", "2025-10-03" - ] + ], + "roll": { + "owners": [ + [ + "molbap", + 1 + ] + ], + "n": 1, + "add": 677, + "del": 1, + "files": 3, + "commits": 7 + } }, { "t": "modular v2 + `transformers` format", "tag": "in review", "s": "open", "prs": [ - [ - 48271, - "2026-08-26", - "draft modular v2 + transformers format", - "OPEN" - ] + { + "n": 48271, + "d": "2026-08-26", + "t": "draft modular v2 + transformers format", + "st": "OPEN", + "by": "ArthurZucker", + "a": 15323, + "r": 3245, + "f": 175, + "c": 21, + "b": "" + } ], "note": null, "links": [ @@ -587,7 +1202,20 @@ const TRACKS = [ "span": [ "2026-08-25", "2026-08-26" - ] + ], + "roll": { + "owners": [ + [ + "ArthurZucker", + 1 + ] + ], + "n": 1, + "add": 15323, + "del": 3245, + "files": 175, + "commits": 21 + } }, { "t": "Visual modular — the Atlas", @@ -605,61 +1233,117 @@ const TRACKS = [ "span": [ "2026-06-02", "2026-07-01" - ] + ], + "roll": { + "owners": [], + "n": 0, + "add": 0, + "del": 0, + "files": 0, + "commits": 0 + } }, { "t": "mlinter — model-structure lint rules", "tag": "57 rules · v0.1.4", "s": "done", "prs": [ - [ - 44174, - "2026-03-12", - "Expand model-structure lint rules with a fast AST-based, ruff-like framework", - "MERGED" - ], - [ - 44939, - "2026-03-24", - "refactor: mlinter as its own package", - "MERGED" - ], - [ - 45475, - "2026-04-20", - "chore(qa): split out mlinter", - "MERGED" - ], - [ - 45585, - "2026-04-23", - "qa: bumped mlinter and allow local override", - "MERGED" - ], - [ - 47172, - "2026-07-09", - "Switch mlinter to 0.1.2", - "MERGED" - ], - [ - 47819, - "2026-08-06", - "Post mlinter findings as inline PR review comments", - "MERGED" - ], - [ - 47918, - "2026-08-18", - "Moving mlinter to 0.1.4", - "MERGED" - ], - [ - 48117, - "2026-08-19", - "Enable mlinter findings artifact for inline PR reviews", - "MERGED" - ] + { + "n": 44174, + "d": "2026-03-12", + "t": "Expand model-structure lint rules with a fast AST-based, ruff-like framework", + "st": "MERGED", + "by": "tarekziade", + "a": 1367, + "r": 24, + "f": 4, + "c": 12, + "b": "This draft expands utils/check_modeling_structure.py into a rule-driven linter for model code, with new checks and tests, while keeping runtime very low. Key features: The checker is intentionally AST-only (no heavy imports/execution), so it stays fast (<3 seconds against all models) It's built in a" + }, + { + "n": 44939, + "d": "2026-03-24", + "t": "refactor: mlinter as its own package", + "st": "MERGED", + "by": "tarekziade", + "a": 2038, + "r": 1446, + "f": 25, + "c": 8, + "b": "Refactored and cleaned up model linter separated package one rule per module refactored legacy checks into their own rules simplified pattern, duplication removal" + }, + { + "n": 45475, + "d": "2026-04-20", + "t": "chore(qa): split out mlinter", + "st": "MERGED", + "by": "tarekziade", + "a": 54, + "r": 3405, + "f": 29, + "c": 10, + "b": "Split out the mlinter tool see https://github.com/huggingface/transformers-mlinter We want to be able to: use it from other CI projects remove the ability to alter the linter from Transformers PRs This change should keep the exact same behavior than before" + }, + { + "n": 45585, + "d": "2026-04-23", + "t": "qa: bumped mlinter and allow local override", + "st": "MERGED", + "by": "tarekziade", + "a": 323, + "r": 28, + "f": 6, + "c": 5, + "b": "bumps mlinter to 0.1.1 add a local config that can be used to quickly override the linter rules" + }, + { + "n": 47172, + "d": "2026-07-09", + "t": "Switch mlinter to 0.1.2", + "st": "MERGED", + "by": "tarekziade", + "a": 165, + "r": 3, + "f": 4, + "c": 4, + "b": "runs on current mlinter 0.1.2 models that fail to pass the linter TRF014 (trust_remote_code) │ auto, legitimate loader exemption we can keep that TRF016 (dead do_* flags) │ kimi_k25, minimax_m3_vl TRF018 (super()._init_weights) | cosmos3_omni, diffusion_gemma, eomt_dinov3, exaone4_5, gemma4, higgs_a" + }, + { + "n": 47819, + "d": "2026-08-06", + "t": "Post mlinter findings as inline PR review comments", + "st": "MERGED", + "by": "ydshieh", + "a": 198, + "r": 6, + "f": 2, + "c": 2, + "b": "What this does Adds a post-mlinter-review job to pr-ci-post-dashboard-link.yml that: Downloads the mlinter-findings artifact produced by the check_code_quality job Posts the findings as a single PR review with inline comments anchored to the diff The review includes a summary table (rule → count) an" + }, + { + "n": 47918, + "d": "2026-08-18", + "t": "Moving mlinter to 0.1.4", + "st": "MERGED", + "by": "tarekziade", + "a": 2249, + "r": 131, + "f": 54, + "c": 13, + "b": "Bumps transformers-mlinter to 0.1.4 to pick up the new rules (TRF020 to TRF057), and resolves everything those rules flag. The base pinned transformers-mlinter==0.1.2, whose utils/rules.toml carried only TRF001–TRF019. This bump adds 37 new rules in the TRF020–TRF057 range, and every finding they re" + }, + { + "n": 48117, + "d": "2026-08-19", + "t": "Enable mlinter findings artifact for inline PR reviews", + "st": "MERGED", + "by": "ydshieh", + "a": 1, + "r": 1, + "f": 1, + "c": 1, + "b": "Summary Last piece of the mlinter inline review pipeline, verified in #48109. Adds --output-json mlinter-findings.json to the mlinter invocation so the check_code_quality job uploads a mlinter-findings artifact. The post-mlinter-review job (already on main) then downloads it and posts inline review " + } ], "note": "Replaces the board’s “coding rules to ruff linter”: it became its own AST-based, ruff-like package rather than ruff rules.", "links": [], @@ -667,7 +1351,24 @@ const TRACKS = [ "span": [ "2026-03-11", "2026-08-19" - ] + ], + "roll": { + "owners": [ + [ + "tarekziade", + 6 + ], + [ + "ydshieh", + 2 + ] + ], + "n": 8, + "add": 6395, + "del": 5044, + "files": 125, + "commits": 55 + } }, { "t": "Transpilation: transformers → MLX", @@ -689,49 +1390,93 @@ const TRACKS = [ "span": [ "2026-04-15", "2026-06-16" - ] + ], + "roll": { + "owners": [], + "n": 0, + "add": 0, + "del": 0, + "files": 0, + "commits": 0 + } }, { "t": "Community integrations docs", "tag": "10 downstream frameworks", "s": "done", "prs": [ - [ - 42932, - "2026-01-07", - "[docs] inference engines", - "MERGED" - ], - [ - 43185, - "2026-01-09", - "[docs] llama.cpp", - "MERGED" - ], - [ - 43176, - "2026-01-13", - "[docs] tensorrt-llm", - "MERGED" - ], - [ - 43294, - "2026-01-20", - "[docs] pretraining", - "MERGED" - ], - [ - 43312, - "2026-01-20", - "[docs] finetuning", - "MERGED" - ], - [ - 43241, - "2026-02-05", - "[docs] deploying", - "MERGED" - ] + { + "n": 42932, + "d": "2026-01-07", + "t": "[docs] inference engines", + "st": "MERGED", + "by": "stevhliu", + "a": 248, + "r": 208, + "f": 5, + "c": 3, + "b": "starts off a new section dedicated to our integrations with inference engines" + }, + { + "n": 43185, + "d": "2026-01-09", + "t": "[docs] llama.cpp", + "st": "MERGED", + "by": "stevhliu", + "a": 72, + "r": 6, + "f": 2, + "c": 1, + "b": "adds llama.cpp guide explaining how it uses transformers to the new community integrations section create separate section for deployment (more to follow soon here!)" + }, + { + "n": 43176, + "d": "2026-01-13", + "t": "[docs] tensorrt-llm", + "st": "MERGED", + "by": "stevhliu", + "a": 49, + "r": 8, + "f": 2, + "c": 2, + "b": "adds tensorrt-llm guide showing how it uses transformers to the new community integrations section" + }, + { + "n": 43294, + "d": "2026-01-20", + "t": "[docs] pretraining", + "st": "MERGED", + "by": "stevhliu", + "a": 106, + "r": 0, + "f": 3, + "c": 3, + "b": "adds ecosystem integration docs for pretraining frameworks (nanotron, torchtitan)" + }, + { + "n": 43312, + "d": "2026-01-20", + "t": "[docs] finetuning", + "st": "MERGED", + "by": "stevhliu", + "a": 196, + "r": 9, + "f": 4, + "c": 4, + "b": "adds ecosystem integration docs for axolotl, trl, and unsloth" + }, + { + "n": 43241, + "d": "2026-02-05", + "t": "[docs] deploying", + "st": "MERGED", + "by": "stevhliu", + "a": 230, + "r": 24, + "f": 8, + "c": 3, + "b": "adds ecosystem integration docs for deploying with Candle, ExecuTorch, and MLX" + } ], "note": null, "links": [], @@ -739,7 +1484,20 @@ const TRACKS = [ "span": [ "2026-01-07", "2026-02-05" - ] + ], + "roll": { + "owners": [ + [ + "stevhliu", + 6 + ] + ], + "n": 6, + "add": 901, + "del": 255, + "files": 24, + "commits": 16 + } } ] }, @@ -753,36 +1511,66 @@ const TRACKS = [ "tag": "vLLM imports it", "s": "done", "prs": [ - [ - 41580, - "2025-11-13", - "Refactor weight loading", - "MERGED" - ], - [ - 42191, - "2025-11-14", - "Much more efficient and clear weight initialization and tie weights", - "MERGED" - ], - [ - 42254, - "2025-11-20", - "Update conversion mapping to separate renaming from converting", - "MERGED" - ], - [ - 42634, - "2025-12-11", - "allow registration of custom checkpoint conversion mappings", - "MERGED" - ], - [ - 44300, - "2026-03-26", - "Dynamic weight conversion is recursive", - "MERGED" - ] + { + "n": 41580, + "d": "2025-11-13", + "t": "Refactor weight loading", + "st": "MERGED", + "by": "ArthurZucker", + "a": 8733, + "r": 7189, + "f": 586, + "c": 387, + "b": "CORE REFACTORING, loading, converting, logging More helpful debugging report when loading weights If you just want to fuse qkv: It can. You just need to make sure you change the model code and pouf! WeightConverter( [\"self_attn.q_proj\", \"self_attn.k_proj\", \"self_attn.v_proj\"], \"self_attn.qkv_proj\", " + }, + { + "n": 42191, + "d": "2025-11-14", + "t": "Much more efficient and clear weight initialization and tie weights", + "st": "MERGED", + "by": "Cyrilvallez", + "a": 3341, + "r": 5443, + "f": 416, + "c": 45, + "b": "Follow-up to #41580. This PR is focused on speed and efficiency, as well as clarity. It avoids the following issues/bottlenecks: No longer need to switch dynamically the class of the Parameters (which is hard to understand and may lead to issues with quantization etc, as it's not always clear if the" + }, + { + "n": 42254, + "d": "2025-11-20", + "t": "Update conversion mapping to separate renaming from converting", + "st": "MERGED", + "by": "ArthurZucker", + "a": 525, + "r": 329, + "f": 5, + "c": 34, + "b": "Update the conversion: to support torchao" + }, + { + "n": 42634, + "d": "2025-12-11", + "t": "allow registration of custom checkpoint conversion mappings", + "st": "MERGED", + "by": "winglian", + "a": 52, + "r": 1, + "f": 2, + "c": 5, + "b": "For custom modeling, this provides a clean route to adding new or overwriting existing mappings with custom mappings. Additionally, it properly memoizes the cached mappings. Before submitting This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case). Did you rea" + }, + { + "n": 44300, + "d": "2026-03-26", + "t": "Dynamic weight conversion is recursive", + "st": "MERGED", + "by": "zucchini-nlp", + "a": 2539, + "r": 470, + "f": 18, + "c": 48, + "b": "The recursive feature is needed for me in #44252 to allow timm backbone define its conversion only once. Also it currently allows to delete \"t5gemma2\" from conversion, allowing its backbones to rename weight" + } ], "note": "vLLM’s transformers backend imports `transformers.conversion_mapping.get_model_conversion_mapping` directly.", "links": [ @@ -795,19 +1583,50 @@ const TRACKS = [ "span": [ "2025-11-03", "2026-03-26" - ] + ], + "roll": { + "owners": [ + [ + "ArthurZucker", + 2 + ], + [ + "Cyrilvallez", + 1 + ], + [ + "winglian", + 1 + ], + [ + "zucchini-nlp", + 1 + ] + ], + "n": 5, + "add": 15190, + "del": 13432, + "files": 1027, + "commits": 519 + } }, { "t": "Model patching API", "tag": "supported monkey-patching", "s": "done", "prs": [ - [ - 43917, - "2026-03-02", - "Model patching API", - "MERGED" - ] + { + "n": 43917, + "d": "2026-03-02", + "t": "Model patching API", + "st": "MERGED", + "by": "IlyasMoutawwakil", + "a": 1157, + "r": 3, + "f": 5, + "c": 27, + "b": "Integrates a library wide monkey patching API to fix #43284 and allow things like restructuring the expert weights and fusing the qkv projections. A concrete and complete example: Details from typing import Unpack import torch import torch.nn as nn from transformers import AutoModelForCausalLM, Conc" + } ], "note": null, "links": [ @@ -820,31 +1639,62 @@ const TRACKS = [ "span": [ "2026-02-11", "2026-03-02" - ] + ], + "roll": { + "owners": [ + [ + "IlyasMoutawwakil", + 1 + ] + ], + "n": 1, + "add": 1157, + "del": 3, + "files": 5, + "commits": 27 + } }, { "t": "`@auto_docstring`", "tag": "docstrings from signatures", "s": "done", "prs": [ - [ - 33771, - "2025-05-08", - "[`AutoDocstring`] Based on inspect parsing of the signature", - "MERGED" - ], - [ - 38242, - "2025-06-23", - "Add support for auto_docstring with model outputs", - "MERGED" - ], - [ - 39011, - "2025-06-26", - "[docs] @auto_docstring", - "MERGED" - ] + { + "n": 33771, + "d": "2025-05-08", + "t": "[`AutoDocstring`] Based on inspect parsing of the signature", + "st": "MERGED", + "by": "ArthurZucker", + "a": 18216, + "r": 46742, + "f": 405, + "c": 125, + "b": "Config stay explicit any place that had non explicit (dynamic docstring) gets replaced we can leverage inheritance for forward dynamic typing I think" + }, + { + "n": 38242, + "d": "2025-06-23", + "t": "Add support for auto_docstring with model outputs", + "st": "MERGED", + "by": "yonigozlan", + "a": 6972, + "r": 8993, + "f": 177, + "c": 21, + "b": "As the title says. However not sure if worth it, as there are a lot of different descriptions currently for the same args. The descriptions that are different from the one in ModelOutputArgs are left in the ModelOutput docstring, which is why the there's not that much more deletions than additions i" + }, + { + "n": 39011, + "d": "2025-06-26", + "t": "[docs] @auto_docstring", + "st": "MERGED", + "by": "stevhliu", + "a": 107, + "r": 103, + "f": 3, + "c": 2, + "b": "Refactors the @auto_docstring docs: adds a link to it from the modular_transformers.md doc to help with documentation when contributing a model reorganizes sections to start with practical examples of how to use the decorator, documenting arguments, and checking the docstrings first remove the emoji" + } ], "note": null, "links": [], @@ -852,37 +1702,82 @@ const TRACKS = [ "span": [ "2024-10-10", "2025-06-26" - ] + ], + "roll": { + "owners": [ + [ + "ArthurZucker", + 1 + ], + [ + "yonigozlan", + 1 + ], + [ + "stevhliu", + 1 + ] + ], + "n": 3, + "add": 25295, + "del": 55838, + "files": 585, + "commits": 148 + } }, { "t": "Pipelines match the Hub spec", "tag": "spec-compliant pipelines", "s": "done", "prs": [ - [ - 33730, - "2024-09-27", - "Make audio classification pipeline spec-compliant and add test", - "MERGED" - ], - [ - 33769, - "2024-10-01", - "Make ASR pipeline compliant with Hub spec + add tests", - "MERGED" - ], - [ - 33899, - "2024-10-08", - "Image pipelines spec compliance", - "MERGED" - ], - [ - 34288, - "2024-10-22", - "Sync video classification pipeline with huggingface_hub spec", - "MERGED" - ] + { + "n": 33730, + "d": "2024-09-27", + "t": "Make audio classification pipeline spec-compliant and add test", + "st": "MERGED", + "by": "Rocketknight1", + "a": 130, + "r": 3, + "f": 3, + "c": 18, + "b": "cc @LysandreJik, this test splits the large PR at #33452 into manageable pieces. This first PR will be the largest, since it includes the testing utils, as well as testing compliance for AudioClassificationPipeline. The other single-pipeline PRs will be much smaller, since they can just import the s" + }, + { + "n": 33769, + "d": "2024-10-01", + "t": "Make ASR pipeline compliant with Hub spec + add tests", + "st": "MERGED", + "by": "Rocketknight1", + "a": 42, + "r": 9, + "f": 5, + "c": 8, + "b": "As well as making the ASR pipeline compliant, this PR also refactors out the output test, so we can just import that in future PRs as well." + }, + { + "n": 33899, + "d": "2024-10-08", + "t": "Image pipelines spec compliance", + "st": "MERGED", + "by": "Rocketknight1", + "a": 153, + "r": 44, + "f": 13, + "c": 15, + "b": "This PR checks for Hub spec compliance for a lot of image pipelines that have very similar inputs. In most cases, no changes are required except deprecating the timeout argument, and renaming the images input to inputs." + }, + { + "n": 34288, + "d": "2024-10-22", + "t": "Sync video classification pipeline with huggingface_hub spec", + "st": "MERGED", + "by": "Rocketknight1", + "a": 61, + "r": 7, + "f": 3, + "c": 2, + "b": "" + } ], "note": null, "links": [], @@ -890,37 +1785,74 @@ const TRACKS = [ "span": [ "2024-09-27", "2024-10-22" - ] + ], + "roll": { + "owners": [ + [ + "Rocketknight1", + 4 + ] + ], + "n": 4, + "add": 386, + "del": 63, + "files": 24, + "commits": 43 + } }, { "t": "Processor / VLM standardization", "tag": "one processor contract", "s": "done", "prs": [ - [ - 37915, - "2025-05-27", - "[transformers x vLLM] standardize processors", - "MERGED" - ], - [ - 41474, - "2025-10-16", - "🚨 [v5] Toggle the serialization format in processors", - "MERGED" - ], - [ - 41633, - "2025-11-07", - "[v5] 🚨Refactor subprocessors handling in processors", - "MERGED" - ], - [ - 41773, - "2025-11-26", - "Simplify and standardize processor tests", - "MERGED" - ] + { + "n": 37915, + "d": "2025-05-27", + "t": "[transformers x vLLM] standardize processors", + "st": "MERGED", + "by": "zucchini-nlp", + "a": 1167, + "r": 79, + "f": 28, + "c": 19, + "b": "Part of #37780. The design was tested on different model types: Oke, I verified that the inference works for all models, unless I forgot about some new ones. Here is the list I tested. A few models (blip2, gotOcr, gemma3) won't be supported in the first release. Gemma3 is already planned after we me" + }, + { + "n": 41474, + "d": "2025-10-16", + "t": "🚨 [v5] Toggle the serialization format in processors", + "st": "MERGED", + "by": "zucchini-nlp", + "a": 137, + "r": 227, + "f": 23, + "c": 11, + "b": "As per title, in v5 we will start saving in the new nested format. The loading is still supported though for old-format configs Also, we don't need an optional_attributes field that does not do what it is called. Since it is only chat template, let's just manually pop and set it. For audio tokenizer" + }, + { + "n": 41633, + "d": "2025-11-07", + "t": "[v5] 🚨Refactor subprocessors handling in processors", + "st": "MERGED", + "by": "yonigozlan", + "a": 368, + "r": 839, + "f": 149, + "c": 29, + "b": "Refactor the handling of subprocessors in processors. Main change is that we deduce the subprocessors from the init signature instead of having to manually add \"subprocessor_class\" attributes. This means we can remove all attributes attribute in processors, along with all \"subprocessor\"_class attrib" + }, + { + "n": 41773, + "d": "2025-11-26", + "t": "Simplify and standardize processor tests", + "st": "MERGED", + "by": "yonigozlan", + "a": 1925, + "r": 5197, + "f": 113, + "c": 52, + "b": "Improve ProcessorTestMixin to standardize processor tests, especially the setup part. Requires #41633" + } ], "note": null, "links": [], @@ -928,7 +1860,24 @@ const TRACKS = [ "span": [ "2025-05-01", "2025-11-26" - ] + ], + "roll": { + "owners": [ + [ + "zucchini-nlp", + 2 + ], + [ + "yonigozlan", + 2 + ] + ], + "n": 4, + "add": 3597, + "del": 6342, + "files": 313, + "commits": 111 + } } ] }, @@ -942,30 +1891,54 @@ const TRACKS = [ "tag": "one API for n-d parallelism", "s": "done", "prs": [ - [ - 37877, - "2025-05-20", - "parallelism goes brrr", - "MERGED" - ], - [ - 38262, - "2025-05-21", - "Protect ParallelInterface", - "MERGED" - ], - [ - 36835, - "2025-05-23", - "Enhance Model Loading By Providing Parallelism, Uses Optional Env Flag", - "MERGED" - ], - [ - 39562, - "2025-07-22", - "Fixes needed for n-d parallelism and TP", - "MERGED" - ] + { + "n": 37877, + "d": "2025-05-20", + "t": "parallelism goes brrr", + "st": "MERGED", + "by": "nmntz", + "a": 1515, + "r": 136, + "f": 6, + "c": 59, + "b": "Fixes # (issue) Before submitting This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case). Did you read the contributor guideline, Pull Request section? Was this discussed/approved via a Github issue or the forum? Please add a link to it if that's the case. Di" + }, + { + "n": 38262, + "d": "2025-05-21", + "t": "Protect ParallelInterface", + "st": "MERGED", + "by": "ArthurZucker", + "a": 22, + "r": 17, + "f": 3, + "c": 2, + "b": "" + }, + { + "n": 36835, + "d": "2025-05-23", + "t": "Enhance Model Loading By Providing Parallelism, Uses Optional Env Flag", + "st": "MERGED", + "by": "inf3rnus", + "a": 234, + "r": 76, + "f": 4, + "c": 34, + "b": "modeling_utils.py has been modified to allow for the parallelized loading of weights. Two env vars have been introduced, HF_ENABLE_PARALLEL_LOADING which allows weights to be loaded in a multiprocessing pool and HF_PARALLEL_LOADING_WORKERS which specifies how many child processes you want to spawn t" + }, + { + "n": 39562, + "d": "2025-07-22", + "t": "Fixes needed for n-d parallelism and TP", + "st": "MERGED", + "by": "winglian", + "a": 3, + "r": 3, + "f": 1, + "c": 2, + "b": "related: #38949 Fixes # (issue) Before submitting This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case). Did you read the contributor guideline, Pull Request section? Was this discussed/approved via a Github issue or the forum? Please add a link to it if tha" + } ], "note": null, "links": [], @@ -973,37 +1946,86 @@ const TRACKS = [ "span": [ "2025-03-19", "2025-07-22" - ] + ], + "roll": { + "owners": [ + [ + "inf3rnus", + 1 + ], + [ + "nmntz", + 1 + ], + [ + "ArthurZucker", + 1 + ], + [ + "winglian", + 1 + ] + ], + "n": 4, + "add": 1774, + "del": 232, + "files": 14, + "commits": 97 + } }, { "t": "Tensor parallel → DTensor", "tag": "enables native FSDP2 + TP", "s": "done", "prs": [ - [ - 34184, - "2024-11-18", - "Simplify Tensor Parallel implementation with PyTorch TP", - "MERGED" - ], - [ - 34800, - "2024-11-19", - "protect tensor parallel usage", - "MERGED" - ], - [ - 34194, - "2025-02-18", - "feat: add support for tensor parallel training workflow with accelerate", - "MERGED" - ], - [ - 47579, - "2026-08-21", - "🚨 TP dtensor API inference + training", - "MERGED" - ] + { + "n": 34184, + "d": "2024-11-18", + "t": "Simplify Tensor Parallel implementation with PyTorch TP", + "st": "MERGED", + "by": "kwen2501", + "a": 357, + "r": 92, + "f": 18, + "c": 26, + "b": "This PR uses the torch.distributed.tensor.parallel subpackage to implement Tensor Parallel for Llama (as an example). The motivation is multi-fold: to make modeling code simple as single-worker case: all manual TP implementations under if self.config.pretraining_tp > 1 can be removed. to make tensor" + }, + { + "n": 34800, + "d": "2024-11-19", + "t": "protect tensor parallel usage", + "st": "MERGED", + "by": "ArthurZucker", + "a": 11, + "r": 5, + "f": 2, + "c": 1, + "b": "Fixes #34795" + }, + { + "n": 34194, + "d": "2025-02-18", + "t": "feat: add support for tensor parallel training workflow with accelerate", + "st": "MERGED", + "by": "kmehant", + "a": 133, + "r": 6, + "f": 8, + "c": 8, + "b": "1. Add apply_tensor_parallel API to apply TP plan to Llama and Granite models Already merged in to HF/transformers. 2. Introduce tp_size user facing argument to be further consumed by accelerate (see huggingface/accelerate#3173) 3. Allows for e2e TP training. Please review in conjunction with huggin" + }, + { + "n": 47579, + "d": "2026-08-21", + "t": "🚨 TP dtensor API inference + training", + "st": "MERGED", + "by": "3outeille", + "a": 1234, + "r": 2161, + "f": 27, + "c": 244, + "b": "Summary This PR replaces the legacy Transformers tensor-parallel implementation with a DTensor-native backend while preserving efficient inference for kernels that operate on local tensors + loading time. We outsource all the placement and parameters sharding logic to Dtensor which will enable us to" + } ], "note": "The DTensor PR is the one the rest of the parallelism stack depends on.", "links": [], @@ -1011,43 +2033,98 @@ const TRACKS = [ "span": [ "2024-10-15", "2026-08-21" - ] + ], + "roll": { + "owners": [ + [ + "kwen2501", + 1 + ], + [ + "ArthurZucker", + 1 + ], + [ + "kmehant", + 1 + ], + [ + "3outeille", + 1 + ] + ], + "n": 4, + "add": 1735, + "del": 2264, + "files": 55, + "commits": 279 + } }, { "t": "Expert parallel (EP)", "tag": "MoE sharding", "s": "done", "prs": [ - [ - 39501, - "2025-07-25", - "Add ep", - "MERGED" - ], - [ - 45473, - "2026-04-21", - "Fix EP: RouterParallel shape, tp_plan property, grouped_mm sentinels", - "MERGED" - ], - [ - 45279, - "2026-04-22", - "add expert parallelism for gemma-4-26B-A4B-it", - "MERGED" - ], - [ - 45436, - "2026-04-22", - "Add expert parallelism (EP) config support for Qwen3 MoE", - "MERGED" - ], - [ - 45621, - "2026-05-04", - "Better Grouped GEMM + EP", - "MERGED" - ] + { + "n": 39501, + "d": "2025-07-25", + "t": "Add ep", + "st": "MERGED", + "by": "ArthurZucker", + "a": 436, + "r": 186, + "f": 9, + "c": 25, + "b": "Add support for expert parallel!" + }, + { + "n": 45473, + "d": "2026-04-21", + "t": "Fix EP: RouterParallel shape, tp_plan property, grouped_mm sentinels", + "st": "MERGED", + "by": "AmineDiro", + "a": 190, + "r": 48, + "f": 5, + "c": 7, + "b": "While benchmarking Qwen3-30B-A3B SFT training with Expert Parallelism (EP) using TRL, I found three bugs that combine to produce silently wrong results or NaN loss. Every existing test uses tp_plan=\"auto\" which bypasses RouterParallel and EP sentinel handling entirely if I understand correctly. Thes" + }, + { + "n": 45279, + "d": "2026-04-22", + "t": "add expert parallelism for gemma-4-26B-A4B-it", + "st": "MERGED", + "by": "sywangyi", + "a": 19, + "r": 3, + "f": 2, + "c": 2, + "b": "Fixes # (issue) Code Agent Policy The Transformers repo is currently being overwhelmed by a large number of PRs and issue comments written by code agents. We are currently bottlenecked by our ability to review and respond to them. As a result, we ask that new users do not submit pure code agent PRs " + }, + { + "n": 45436, + "d": "2026-04-22", + "t": "Add expert parallelism (EP) config support for Qwen3 MoE", + "st": "MERGED", + "by": "AmineDiro", + "a": 35, + "r": 0, + "f": 5, + "c": 11, + "b": "Summary Added base_model_ep_plan to Qwen3MoeConfig enabling expert parallelism via DistributedConfig(enable_expert_parallel=True) Depends on #45473 Test plan Tested on 8×H100 with torchrun --nproc_per_node=8 using Qwen/Qwen3-30B-A3B (128 experts, 4 KV heads): TP=2, DP=4: forward ✓, backward ✓, exper" + }, + { + "n": 45621, + "d": "2026-05-04", + "t": "Better Grouped GEMM + EP", + "st": "MERGED", + "by": "IlyasMoutawwakil", + "a": 394, + "r": 219, + "f": 5, + "c": 36, + "b": "The idea is that we shouldn't be clamping experts ids at all, clamping them makes their tokens get prjected as if they were routed to the last expert, instead, we should let the sentinels be: moved to the queue by the sorting (num_experts/sentinel is the biggest value) dropped by the histogram (beca" + } ], "note": null, "links": [], @@ -1055,31 +2132,74 @@ const TRACKS = [ "span": [ "2025-07-17", "2026-05-04" - ] + ], + "roll": { + "owners": [ + [ + "AmineDiro", + 2 + ], + [ + "ArthurZucker", + 1 + ], + [ + "sywangyi", + 1 + ], + [ + "IlyasMoutawwakil", + 1 + ] + ], + "n": 5, + "add": 1074, + "del": 456, + "files": 26, + "commits": 81 + } }, { "t": "Pipeline parallel (PP)", "tag": "plan + naive inference", "s": "done", "prs": [ - [ - 36091, - "2025-02-12", - "Add pipeline parallel plan to `PretrainedConfig` and `PreTrainedModel`", - "MERGED" - ], - [ - 44699, - "2026-03-20", - "Fix several based models' pipeline parallel support", - "MERGED" - ], - [ - 47289, - "2026-08-25", - "Pipeline parallel naive inference", - "MERGED" - ] + { + "n": 36091, + "d": "2025-02-12", + "t": "Add pipeline parallel plan to `PretrainedConfig` and `PreTrainedModel`", + "st": "MERGED", + "by": "hmellor", + "a": 188, + "r": 0, + "f": 50, + "c": 25, + "b": "Fixes # (issue) Before submitting This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case). Did you read the contributor guideline, Pull Request section? Was this discussed/approved via a Github issue or the forum? Please add a link to it if that's the case. Di" + }, + { + "n": 44699, + "d": "2026-03-20", + "t": "Fix several based models' pipeline parallel support", + "st": "MERGED", + "by": "hmellor", + "a": 219, + "r": 148, + "f": 52, + "c": 6, + "b": "These models have base_model_pp_plans but currently do not work because the base model's forward pass depends on all the layers being Qwen2VLDecoderLayer. i.e. if one of the layers is removed/replaced with Identity, decoder_layer.attention_type will fail with an attribute error. This PR fixes the is" + }, + { + "n": 47289, + "d": "2026-08-25", + "t": "Pipeline parallel naive inference", + "st": "MERGED", + "by": "3outeille", + "a": 758, + "r": 10, + "f": 9, + "c": 317, + "b": "Summary Adding a very naive pipeline parallel inference engine as a baseline (no specific schedule such as gpipe, 1f1b yet) supports tied/untied weight embeddings To make it work seamless with generate(), we are broadcasting every logits to every ranks. In long term, we want only the last stage to h" + } ], "note": null, "links": [], @@ -1087,25 +2207,54 @@ const TRACKS = [ "span": [ "2025-02-07", "2026-08-25" - ] + ], + "roll": { + "owners": [ + [ + "hmellor", + 2 + ], + [ + "3outeille", + 1 + ] + ], + "n": 3, + "add": 1165, + "del": 158, + "files": 111, + "commits": 348 + } }, { "t": "Context parallel (CP)", "tag": "long-context sharding", "s": "done", "prs": [ - [ - 40619, - "2025-09-03", - "[CP] Add attention_mask to the buffer when the mask is causal ", - "MERGED" - ], - [ - 48167, - "2026-08-21", - "Apply context parallelism to the evaluation path", - "MERGED" - ] + { + "n": 40619, + "d": "2025-09-03", + "t": "[CP] Add attention_mask to the buffer when the mask is causal ", + "st": "MERGED", + "by": "kashif", + "a": 25, + "r": 14, + "f": 1, + "c": 3, + "b": "Ensured that the attention_mask is always checked for causality before being added to the buffer, and that this validation is performed only once for performance. Modified the logic so that the attention_mask is appended to buffers only after successful validation, preventing non-causal masks from b" + }, + { + "n": 48167, + "d": "2026-08-21", + "t": "Apply context parallelism to the evaluation path", + "st": "MERGED", + "by": "qgallouedec", + "a": 4, + "r": 2, + "f": 1, + "c": 2, + "b": "With context parallelism enabled, eval_loss is reported as cp_size times its true value. Training loss is correct; only evaluation is wrong. Same everything, varying only cp_size: Run it once per cp_size (no config file needed): FSDP=\"--use_fsdp --fsdp_version 2 --fsdp_auto_wrap_policy TRANSFORMER_B" + } ], "note": null, "links": [], @@ -1113,31 +2262,66 @@ const TRACKS = [ "span": [ "2025-09-02", "2026-08-21" - ] + ], + "roll": { + "owners": [ + [ + "kashif", + 1 + ], + [ + "qgallouedec", + 1 + ] + ], + "n": 2, + "add": 29, + "del": 16, + "files": 2, + "commits": 5 + } }, { "t": "FSDP2 + DTensor in Trainer", "tag": "end-to-end FSDP CI", "s": "done", "prs": [ - [ - 42436, - "2025-12-03", - "Fix mixed torch.Tensor and DTensor in generate when use FSDP2 + LoRA", - "MERGED" - ], - [ - 46246, - "2026-05-28", - "[`Revert`] FSDP+Dtensor refactor related changes", - "MERGED" - ], - [ - 47357, - "2026-07-23", - "Add FSDP CI and end-to-end FSDP tests + save fsdp", - "MERGED" - ] + { + "n": 42436, + "d": "2025-12-03", + "t": "Fix mixed torch.Tensor and DTensor in generate when use FSDP2 + LoRA", + "st": "MERGED", + "by": "Xiao-Chenguang", + "a": 2, + "r": 0, + "f": 1, + "c": 3, + "b": "model.generate raise an Fix mixed torch.Tensor and DTensor when FSDP2 and LoRA is applied. The trainer class is the base class for trl trainers like PPOTrainer and GRPOTrainer in which model.generate is called during the training loop. When FSDP2 is used, all gather and reshard is automatically done" + }, + { + "n": 46246, + "d": "2026-05-28", + "t": "[`Revert`] FSDP+Dtensor refactor related changes", + "st": "MERGED", + "by": "vasqu", + "a": 3011, + "r": 6332, + "f": 286, + "c": 8, + "b": "As per title 👀 Discussed internally" + }, + { + "n": 47357, + "d": "2026-07-23", + "t": "Add FSDP CI and end-to-end FSDP tests + save fsdp", + "st": "MERGED", + "by": "3outeille", + "a": 960, + "r": 82, + "f": 16, + "c": 40, + "b": "Summary Part of FSDP orchestration stack (2/2). Replaces #46990. #47352 (1/2) → this PR (2/2) Add FSDP DCP (distributed_checkpoint=True) paths in save_pretrained Add FSDP gather into single checkpoint saving Add FSDPTesterMixin and tests/test_fsdp_mixin.py (load/save/generation) Add base_model_fsdp_" + } ], "note": null, "links": [], @@ -1145,31 +2329,70 @@ const TRACKS = [ "span": [ "2025-12-03", "2026-07-23" - ] + ], + "roll": { + "owners": [ + [ + "Xiao-Chenguang", + 1 + ], + [ + "vasqu", + 1 + ], + [ + "3outeille", + 1 + ] + ], + "n": 3, + "add": 3973, + "del": 6414, + "files": 303, + "commits": 51 + } }, { "t": "Trainer refactor", "tag": "unbloat + modular Trainer", "s": "wip", "prs": [ - [ - 35567, - "2025-01-29", - "Trainer Refactor: Part 1", - "MERGED" - ], - [ - 38090, - "2025-05-19", - "Refactor `get_XXX_dataloader` from Trainer", - "MERGED" - ], - [ - 43807, - "2026-02-10", - "Refactor trainer init", - "MERGED" - ] + { + "n": 35567, + "d": "2025-01-29", + "t": "Trainer Refactor: Part 1", + "st": "MERGED", + "by": "muellerzr", + "a": 147, + "r": 132, + "f": 3, + "c": 15, + "b": "This is the first chunk in my attempt to clean up the trainer without breaking core capabilities (and integration capabilities/projects). The aim is for readability rather than enhancement of features. I've started this process by: Refactoring how we get the \"core\" training values (max_steps, update" + }, + { + "n": 38090, + "d": "2025-05-19", + "t": "Refactor `get_XXX_dataloader` from Trainer", + "st": "MERGED", + "by": "yaswanth19", + "a": 72, + "r": 80, + "f": 1, + "c": 3, + "b": "Refactors dataloader functions in Trainer" + }, + { + "n": 43807, + "d": "2026-02-10", + "t": "Refactor trainer init", + "st": "MERGED", + "by": "SunMarc", + "a": 319, + "r": 250, + "f": 5, + "c": 16, + "b": "This PR simplifies Trainer __init__: Quantization validation extracted PEFT unwrapping deduplicated Liger Kernel extracted — apply_liger_kernel Label smoother simplified Validations grouped — _validate_args() method consolidates three scattered validation blocks (arg checks, optimizer checks, datase" + } ], "note": null, "links": [], @@ -1177,25 +2400,58 @@ const TRACKS = [ "span": [ "2025-01-07", "2026-02-10" - ] + ], + "roll": { + "owners": [ + [ + "muellerzr", + 1 + ], + [ + "yaswanth19", + 1 + ], + [ + "SunMarc", + 1 + ] + ], + "n": 3, + "add": 538, + "del": 462, + "files": 9, + "commits": 34 + } }, { "t": "Trackio integration", "tag": "training telemetry", "s": "done", "prs": [ - [ - 38814, - "2025-07-22", - "🎯 Trackio integration", - "MERGED" - ], - [ - 39595, - "2025-07-23", - "[Trackio] Allow single-gpu training and monitor power", - "MERGED" - ] + { + "n": 38814, + "d": "2025-07-22", + "t": "🎯 Trackio integration", + "st": "MERGED", + "by": "qgallouedec", + "a": 150, + "r": 9, + "f": 7, + "c": 13, + "b": "from trl import SFTTrainer, SFTConfig from datasets import load_dataset trainer = SFTTrainer( model=\"Qwen/Qwen2.5-0.5B\", args=SFTConfig(report_to=\"trackio\"), train_dataset=load_dataset(\"trl-lib/Capybara\", split=\"train\"), ) trainer.train() Before submitting This PR fixes a typo or improves the docs (" + }, + { + "n": 39595, + "d": "2025-07-23", + "t": "[Trackio] Allow single-gpu training and monitor power", + "st": "MERGED", + "by": "qgallouedec", + "a": 9, + "r": 6, + "f": 1, + "c": 1, + "b": "" + } ], "note": null, "links": [], @@ -1203,7 +2459,20 @@ const TRACKS = [ "span": [ "2025-06-13", "2025-07-23" - ] + ], + "roll": { + "owners": [ + [ + "qgallouedec", + 2 + ] + ], + "n": 2, + "add": 159, + "del": 15, + "files": 8, + "commits": 14 + } }, { "t": "TorchTitan integration", @@ -1213,7 +2482,15 @@ const TRACKS = [ "note": "No PRs found in transformers as of Sep 2026.", "links": [], "d": null, - "span": null + "span": null, + "roll": { + "owners": [], + "n": 0, + "add": 0, + "del": 0, + "files": 0, + "commits": 0 + } }, { "t": "Native sharding via safetensors", @@ -1223,7 +2500,352 @@ const TRACKS = [ "note": "No PRs found in transformers as of Sep 2026.", "links": [], "d": null, - "span": null + "span": null, + "roll": { + "owners": [], + "n": 0, + "add": 0, + "del": 0, + "files": 0, + "commits": 0 + } + } + ] + }, + { + "name": "Export & on-device", + "hue": 190, + "blurb": "Get a transformers model out of Python and onto a device.", + "items": [ + { + "t": "torch.export-compatible modeling", + "tag": "models export cleanly", + "s": "done", + "prs": [ + { + "n": 32168, + "d": "2024-07-29", + "t": "Make static cache compatible with torch.export", + "st": "MERGED", + "by": "guangy10", + "a": 80, + "r": 10, + "f": 2, + "c": 2, + "b": "This PR is to make the static cache impl compatible with both torch.compile() and torch.export() (new). Making a model exportable so that it can be further lowered and run on edge devices through ExecuTorch, which is a native PyTorch framework for on-device inference. An example of lower Gemma-2b to" + }, + { + "n": 33707, + "d": "2024-10-11", + "t": "Generate using exported model and enable gemma2-2b in ExecuTorch", + "st": "MERGED", + "by": "guangy10", + "a": 168, + "r": 0, + "f": 3, + "c": 5, + "b": "Adding generate support for exported model. Adding gemma2-2b to ExecuTorch with tests. Adding an integration test for gemma-2b that we've enabled already. Additional Test in ExecuTorch Running gemma2-2b E2E: cmake-out/examples/models/llama2/llama_main --tokenizer_path=tokenizer_gemma2.bin --model_pa" + }, + { + "n": 35124, + "d": "2025-02-11", + "t": "Add common test for `torch.export` and fix some vision models", + "st": "MERGED", + "by": "qubvel", + "a": 305, + "r": 151, + "f": 77, + "c": 46, + "b": "Add a common slow test to check if a model can be exported with no issues using torch.export.export Add an optional test, to enable it please set test_torch_exportable = True flag for model-specific test. Enable test for vision and video models Fix most of the vision models The main fixes include: U" + }, + { + "n": 37124, + "d": "2025-04-01", + "t": "Make canine model exportable by removing unncessary complicated logic", + "st": "MERGED", + "by": "tugsbayasgalan", + "a": 2, + "r": 2, + "f": 1, + "c": 2, + "b": "In this code, char_seq_length is guaranteed to be an integer, so there is no need to cast it to tensor. By removing this, canine model is now exportable. Fixes # (issue) Before submitting This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case). Did you read th" + }, + { + "n": 37728, + "d": "2025-04-28", + "t": "Gemma3 is Torch Exportable", + "st": "MERGED", + "by": "guangy10", + "a": 369, + "r": 10, + "f": 9, + "c": 2, + "b": "Initial effort to add torch.expport support for the Gemma3 model! Gemma3 provides a 1b variant that is suitable for ExecuTorch to bring it for on-device use-case. This PR is focusing on creating the export recipe and validate the exported model can produce same output as eager. Expand support to oth" + }, + { + "n": 39836, + "d": "2025-08-07", + "t": "Support input_embeds in torch exportable decoders", + "st": "MERGED", + "by": "jackzhxng", + "a": 324, + "r": 84, + "f": 11, + "c": 13, + "b": "Allows specifying inputs_embeds in TorchExportableModule's in order to support export of multimodal model's text decoders. Adds config and generation_config to the constructors to support multimodal models since the TorchExportableModule will wrap the nested text decoder model, which doesn't have it" + }, + { + "n": 47096, + "d": "2026-07-31", + "t": "Exportable kimi", + "st": "MERGED", + "by": "IlyasMoutawwakil", + "a": 517, + "r": 293, + "f": 19, + "c": 13, + "b": "Fixes # (issue) Code Agent Policy The Transformers repo is currently being overwhelmed by a large number of PRs and issue comments written by code agents. We are currently bottlenecked by our ability to review and respond to them. As a result, we ask that new users do not submit pure code agent PRs " + }, + { + "n": 47871, + "d": "2026-08-12", + "t": "Make muse glimmer exportable", + "st": "MERGED", + "by": "IlyasMoutawwakil", + "a": 160, + "r": 286, + "f": 5, + "c": 3, + "b": "Fixes # (issue) Code Agent Policy The Transformers repo is currently being overwhelmed by a large number of PRs and issue comments written by code agents. These often are low-quality, or fix extremely minor issues that occur rarely or never in practice. As a result, we're instituting a rule that fir" + } + ], + "note": null, + "links": [], + "d": "Static cache made export-safe, generate from an exported model, then a common `torch.export` test so every new model stays exportable — plus a long tail of per-model fixes.", + "span": [ + "2024-07-26", + "2026-08-12" + ], + "roll": { + "owners": [ + [ + "guangy10", + 3 + ], + [ + "IlyasMoutawwakil", + 2 + ], + [ + "qubvel", + 1 + ], + [ + "tugsbayasgalan", + 1 + ] + ], + "n": 8, + "add": 1925, + "del": 836, + "files": 127, + "commits": 86 + } + }, + { + "t": "`transformers.exporters`", + "tag": "one export entry point", + "s": "done", + "prs": [ + { + "n": 41992, + "d": "2026-07-02", + "t": "[PoC] HF exporters", + "st": "MERGED", + "by": "IlyasMoutawwakil", + "a": 5468, + "r": 264, + "f": 89, + "c": 406, + "b": "HfExporters: Native, Unified export for PyTorch / ONNX / ExecuTorch Edit: pieces of this PR were merged separately (#42697, #42317), so this one is now focused on the HfExporters machinery itself. What this adds A native, in-Transformers export pipeline — one base class (HfExporter), three subclasse" + }, + { + "n": 47711, + "d": "2026-08-03", + "t": "Fix ImportError in transformers.exporters on torch < 2.8", + "st": "MERGED", + "by": "Neal006", + "a": 1, + "r": 1, + "f": 1, + "c": 2, + "b": "Fixes #47710. import transformers.exporters raises ImportError on torch 2.5, 2.6 and 2.7, all inside the declared torch>=2.5 support range. exporter_executorch.py imports guard_or_true at module scope, and that symbol was added in torch 2.8 where it replaced definitely_true. is_torch_available() che" + }, + { + "n": 47243, + "d": "2026-08-03", + "t": "Executorch exporter fixes", + "st": "MERGED", + "by": "IlyasMoutawwakil", + "a": 1894, + "r": 314, + "f": 28, + "c": 41, + "b": "Been hunting them since we introduced their skips in the original HF exporters PR. Also adds static cache export support, for in-place cache modification with cache position. Also adds multi-token decoding support, so that the decode graph can do prefill (empty cache), single token decode and multi-" + }, + { + "n": 47900, + "d": "2026-08-21", + "t": "Fix DynamicCache reconstruction during ExecuTorch export", + "st": "MERGED", + "by": "eladsegal", + "a": 97, + "r": 25, + "f": 3, + "c": 19, + "b": "Fixes incorrect cache-update graphs when exporting models with DynamicCache to ExecuTorch. torch.export uses the registered pytree functions to flatten and reconstruct DynamicCache during export. The current pytree registration keeps populated key/value tensors, but loses each layer's type and windo" + } + ], + "note": null, + "links": [], + "d": "The HF exporters module: a single supported path out of transformers instead of per-model export scripts.", + "span": [ + "2026-03-19", + "2026-08-21" + ], + "roll": { + "owners": [ + [ + "IlyasMoutawwakil", + 2 + ], + [ + "Neal006", + 1 + ], + [ + "eladsegal", + 1 + ] + ], + "n": 4, + "add": 7460, + "del": 604, + "files": 121, + "commits": 468 + } + }, + { + "t": "ExecuTorch integration", + "tag": "on-device runtime", + "s": "done", + "prs": [ + { + "n": 34101, + "d": "2024-10-17", + "t": "Llama3 and Llama2 are ExecuTorch compatible", + "st": "MERGED", + "by": "guangy10", + "a": 69, + "r": 0, + "f": 1, + "c": 1, + "b": "Llama 2&# is compatible with ExecuTorch. Note that LLama2&3 in ExecuTorch repo has been fully optimized for SOTA perf using its own model definition and optimization. You can read details in https://github.com/pytorch/executorch/tree/main/examples/models/llama2. The work here is to make the Llama mo" + }, + { + "n": 34102, + "d": "2024-10-22", + "t": "Qwen2.5 is ExecuTorch Compatible", + "st": "MERGED", + "by": "guangy10", + "a": 55, + "r": 0, + "f": 1, + "c": 1, + "b": "Qwen2.5 is compatible with ExecuTorch. Before submitting Did you read the contributor guideline, Pull Request section? Was this discussed/approved via a Github issue or the forum? Please add a link to it if that's the case. #33833 Did you write any new necessary tests? Who can review? Anyone in the " + }, + { + "n": 36486, + "d": "2025-03-31", + "t": "Export T5 (encoder-decoder) to ExecuTorch", + "st": "MERGED", + "by": "guangy10", + "a": 324, + "r": 0, + "f": 2, + "c": 1, + "b": "This PR enables exporting T5 model to ExecuTorch, which has been asked by many OSS users. We will need to export T5 encoder and decoder separately (i.e. to separate .pte files when lowering to ExecuTorch), and compose the encoder-decoder for specific task (e.g. summarization) in the ExecuTorch runti" + }, + { + "n": 38646, + "d": "2025-06-13", + "t": "Unbreak optimum-executorch", + "st": "MERGED", + "by": "guangy10", + "a": 64, + "r": 39, + "f": 9, + "c": 3, + "b": "Revert minimal changes made from #37866 that breaks export to ExecuTorch in huggingface/optimum-executorch when developing from latest transformers trunk TODO: Will update with tests shortly Before submitting Did you read the contributor guideline, Pull Request section? Was this discussed/approved v" + }, + { + "n": 41627, + "d": "2025-10-16", + "t": "[`Executorch`] Simplify for encoder models", + "st": "MERGED", + "by": "vasqu", + "a": 25, + "r": 189, + "f": 6, + "c": 2, + "b": "Now that we include a fast path using no vmapping, we can revert the extra treatment. Followup to #41586" + }, + { + "n": 47201, + "d": "2026-07-09", + "t": "Make executorch exporter tests always use xnnpack backend", + "st": "MERGED", + "by": "tarekziade", + "a": 19, + "r": 6, + "f": 3, + "c": 19, + "b": "The ExecuTorch export tests target CPU (via xnnpack on CPU), and its deps now live in the CPU torch-light CI image. Mark test_executorch_export with a new require_non_gpu decorator so they run on the CPU image, not the GPU nightly. Lower DynamoExporter.min_versions torch 2.11.0 -> 2.10.0 so the test" + } + ], + "note": "ExecuTorch was pulled from CI in #47242 until the latest version is supported — the integration ships, the CI coverage is currently paused.", + "links": [ + [ + "ExecuTorch docs", + "https://huggingface.co/docs/transformers/community_integrations/executorch" + ] + ], + "d": "Llama, Qwen2.5, T5 encoder-decoder and the encoder models exported to ExecuTorch; xnnpack backend in the exporter tests.", + "span": [ + "2024-10-11", + "2026-07-09" + ], + "roll": { + "owners": [ + [ + "guangy10", + 4 + ], + [ + "vasqu", + 1 + ], + [ + "tarekziade", + 1 + ] + ], + "n": 6, + "add": 556, + "del": 234, + "files": 22, + "commits": 27 + } } ] }, @@ -1237,50 +2859,165 @@ const TRACKS = [ "tag": "quant via WeightConverter", "s": "done", "prs": [ - [ - 42254, - "2025-11-20", - "Update conversion mapping to separate renaming from converting", - "MERGED" - ], - [ - 42634, - "2025-12-11", - "allow registration of custom checkpoint conversion mappings", - "MERGED" - ] + { + "n": 41138, + "d": "2025-09-29", + "t": "Make quantizers good citizens loading-wise", + "st": "MERGED", + "by": "Cyrilvallez", + "a": 356, + "r": 538, + "f": 15, + "c": 21, + "b": "The quantizers should NEVER need to access the state_dict during param_needs_quantization or create_quantized_param. This is mostly for the following reasons: Most models have a lot of different state dict (due to restricting the size of each file to 5GB), so relying on other params of the state dic" + }, + { + "n": 42043, + "d": "2025-11-18", + "t": "Fix bnb for the weights refactor", + "st": "MERGED", + "by": "SunMarc", + "a": 627, + "r": 185, + "f": 13, + "c": 396, + "b": "This PR fixes bnb support (8bit + 4bit) in the new weight loading logic. Testing from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig model_name = \"meta-llama/Llama-3.2-3B-Instruct\" quantization_config = BitsAndBytesConfig(load_in_4bit=True, bnb_4bit_quant_type=\"nf4\") #mo" + }, + { + "n": 42254, + "d": "2025-11-20", + "t": "Update conversion mapping to separate renaming from converting", + "st": "MERGED", + "by": "ArthurZucker", + "a": 525, + "r": 329, + "f": 5, + "c": 34, + "b": "Update the conversion: to support torchao" + }, + { + "n": 42426, + "d": "2025-11-26", + "t": "Clean bnb integration using weight converter", + "st": "MERGED", + "by": "SunMarc", + "a": 141, + "r": 104, + "f": 4, + "c": 6, + "b": "This PR cleans bnb integration, we don't need to deal with missing keys." + }, + { + "n": 42634, + "d": "2025-12-11", + "t": "allow registration of custom checkpoint conversion mappings", + "st": "MERGED", + "by": "winglian", + "a": 52, + "r": 1, + "f": 2, + "c": 5, + "b": "For custom modeling, this provides a clean route to adding new or overwriting existing mappings with custom mappings. Additionally, it properly memoizes the cached mappings. Before submitting This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case). Did you rea" + }, + { + "n": 44576, + "d": "2026-03-16", + "t": "Disable async loading when quantizing on the fly", + "st": "MERGED", + "by": "SunMarc", + "a": 16, + "r": 5, + "f": 1, + "c": 11, + "b": "Fixes #44387. This PR disable async loading when we want to quantize the model. it is actually faster than doing a semaphore. If a quantizer happens to quantize faster than materializing the weights, we might consider adding adding a semaphore for that. For future reference: I tried to Materialize a" + }, + { + "n": 46325, + "d": "2026-06-08", + "t": "feat(quantizers): support non-weight param names in TorchAo safetensors loading", + "st": "MERGED", + "by": "agesf", + "a": 52, + "r": 17, + "f": 2, + "c": 4, + "b": "Add _discover_quantized_param_names() to dynamically extract parameter names from safetensors metadata instead of hardcoding 'weight' Add aten.normal_ compat patch for TorchAO tensor subclasses What does this PR do? Allows loading quantized checkpoints that have non-standard parameter names." + } ], "note": null, "links": [], - "d": "bnb integration cleaned up to go through the weight converter — the “simplified quantization” item.", + "d": "Quantizers made “good citizens loading-wise”, bnb re-plumbed through the weight converter, custom conversion mappings, async loading disabled when quantizing on the fly, TorchAo safetensors params.", "span": [ - "2025-11-18", - "2025-12-11" - ] + "2025-09-25", + "2026-06-08" + ], + "roll": { + "owners": [ + [ + "SunMarc", + 3 + ], + [ + "Cyrilvallez", + 1 + ], + [ + "ArthurZucker", + 1 + ], + [ + "winglian", + 1 + ] + ], + "n": 7, + "add": 1769, + "del": 1179, + "files": 42, + "commits": 477 + } }, { "t": "Kernel optimization for quant methods", "tag": "FP8 / MXFP4 via `kernels`", "s": "done", "prs": [ - [ - 41897, - "2025-11-04", - "[FPQuant] MXFP8 and MXFP4 backwards support", - "MERGED" - ], - [ - 45634, - "2026-06-02", - "DeepGEMM BF16 + mixed FP8/FP4 + MegaMoE + refactor", - "MERGED" - ], - [ - 47863, - "2026-08-17", - "Bump default flash-attn2 hub kernel version to v3", - "MERGED" - ] + { + "n": 41897, + "d": "2025-11-04", + "t": "[FPQuant] MXFP8 and MXFP4 backwards support", + "st": "MERGED", + "by": "BlackSamorez", + "a": 20, + "r": 5, + "f": 5, + "c": 6, + "b": "This PR adds MXFP4 and MXFP8 backwards support in combination with MXFP4 forward, allowing for lightning-fast QAT on Blackwell GPUs. It's blocked by QuTLASS v0.2.0 and fp_quant v0.3.0 release we're hoping to release within a day or two. Before submitting This PR fixes a typo or improves the docs (yo" + }, + { + "n": 45634, + "d": "2026-06-02", + "t": "DeepGEMM BF16 + mixed FP8/FP4 + MegaMoE + refactor", + "st": "MERGED", + "by": "IlyasMoutawwakil", + "a": 1627, + "r": 486, + "f": 16, + "c": 164, + "b": "Based on #45621 GroupedLinear co-authored-by @sywangyi Code Agent Policy The Transformers repo is currently being overwhelmed by a large number of PRs and issue comments written by code agents. We are currently bottlenecked by our ability to review and respond to them. As a result, we ask that new u" + }, + { + "n": 47863, + "d": "2026-08-17", + "t": "Bump default flash-attn2 hub kernel version to v3", + "st": "MERGED", + "by": "jiqing-feng", + "a": 19, + "r": 3, + "f": 1, + "c": 6, + "b": "What load_and_register_attn_kernel hardcodes version=1 as the default major version for every attention kernel loaded from the Hub. Pin kernels-community/flash-attn2 to v3 through a small per-repo table instead. Why The v1 branch of kernels-community/flash-attn2 only ships builds up to Torch 2.12, s" + } ], "note": null, "links": [], @@ -1288,19 +3025,46 @@ const TRACKS = [ "span": [ "2025-10-27", "2026-08-17" - ] + ], + "roll": { + "owners": [ + [ + "IlyasMoutawwakil", + 1 + ], + [ + "BlackSamorez", + 1 + ], + [ + "jiqing-feng", + 1 + ] + ], + "n": 3, + "add": 1666, + "del": 494, + "files": 22, + "commits": 176 + } }, { "t": "`transformers serve` + quantization", "tag": "quantized serving", "s": "done", "prs": [ - [ - 41253, - "2025-10-23", - "transformers serve quantization docs + some api fixes for bitsandbytes", - "MERGED" - ] + { + "n": 41253, + "d": "2025-10-23", + "t": "transformers serve quantization docs + some api fixes for bitsandbytes", + "st": "MERGED", + "by": "SunMarc", + "a": 39, + "r": 31, + "f": 2, + "c": 18, + "b": "This PR adds some docs for transformers serve x quantization and fixes the quantization api for bitsandbytes. Following PR will add more support for other methods and I will add a section in each quantization method to show what is the preferred setup (compile ? dtype ?) when using serve !" + } ], "note": null, "links": [], @@ -1308,51 +3072,173 @@ const TRACKS = [ "span": [ "2025-10-01", "2025-10-23" - ] + ], + "roll": { + "owners": [ + [ + "SunMarc", + 1 + ] + ], + "n": 1, + "add": 39, + "del": 31, + "files": 2, + "commits": 18 + } }, { - "t": "MacOS / GGUF loading", - "tag": "GGUF checkpoints load", - "s": "done", + "t": "MacOS first-class: GGUF + Metal kernels", + "tag": "Metal dequant via kernels-community", + "s": "open", "prs": [ - [ - 38645, - "2025-07-15", - "support loading qwen3 gguf", - "MERGED" - ], - [ - 39638, - "2025-07-29", - "Support loading Qwen3 MoE GGUF", - "MERGED" - ], - [ - 41111, - "2025-09-24", - "Support loading LFM2 GGUF", - "MERGED" - ] + { + "n": 38645, + "d": "2025-07-15", + "t": "support loading qwen3 gguf", + "st": "MERGED", + "by": "44670", + "a": 31, + "r": 0, + "f": 2, + "c": 6, + "b": "This PR adds gguf loading support for qwen3 models. Before submitting [No] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case). [Yes] Did you read the contributor guideline, Pull Request section? [No afaik] Was this discussed/approved via a Github issue or" + }, + { + "n": 39638, + "d": "2025-07-29", + "t": "Support loading Qwen3 MoE GGUF", + "st": "MERGED", + "by": "ctcanbol", + "a": 31, + "r": 0, + "f": 2, + "c": 7, + "b": "Currently, GGUF versions of Qwen3 MoE models raises \"GGUF model with architecture qwen3moe is not supported yet\" error. This PR resolves this issue. Fixes #39721 Before submitting This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case). Did you read the contri" + }, + { + "n": 41111, + "d": "2025-09-24", + "t": "Support loading LFM2 GGUF", + "st": "MERGED", + "by": "HaroldBenoit", + "a": 57, + "r": 0, + "f": 3, + "c": 6, + "b": "Currently, GGUF versions of LFM2 models raises \"GGUF model with architecture lfm2 is not supported yet\" error. This PR resolves this issue. Before submitting This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case). Did you read the contributor guideline, Pull " + }, + { + "n": 44348, + "d": "2026-02-28", + "t": "Enable MetalConfig to load pre-quantized MLX models from HuggingFace Hub", + "st": "OPEN", + "by": "n0kovo", + "a": 341, + "r": 32, + "f": 4, + "c": 4, + "b": "Summary Most quantized models for Apple Silicon on the Hub are in MLX format. The MetalConfig quantization backend supports on-the-fly quantization of standard checkpoints but cannot load pre-quantized MLX models. This PR fixes the five issues blocking that: quantizers/auto.py: Detect MLX affine qua" + }, + { + "n": 45975, + "d": "2026-05-14", + "t": "GGUF: optional Metal dequant fast path via kernels-community", + "st": "OPEN", + "by": "ArthurZucker", + "a": 127, + "r": 6, + "f": 3, + "c": 3, + "b": "Summary Adds an optional Metal dequant fast path in GGUFDequantize. When the kernels package is installed and kernels-community/gguf-dequant is reachable, MPS GGUF loads (Q4_0 / Q8_0 / Q4_K / Q5_K / Q6_K / IQ4_NL / IQ4_XS) route through one Metal compute kernel per tensor — 2.3–7.5× faster than the " + }, + { + "n": 46357, + "d": "2026-06-02", + "t": "[docs] gguf mps", + "st": "OPEN", + "by": "stevhliu", + "a": 71, + "r": 0, + "f": 1, + "c": 2, + "b": "adds docs for: #45975, dequant fast path through a Metal kernel #45977, native quantized and on-the-fly quantization inference on a Metal kernel. i don't think the docs here overlap with the existing docs in your PR :)" + }, + { + "n": 46792, + "d": "2026-06-26", + "t": "Enable CB paged decode fast path for MPS hub kernels + sensible block_size default", + "st": "OPEN", + "by": "ArthurZucker", + "a": 52, + "r": 7, + "f": 4, + "c": 3, + "b": "Summary Three changes to make continuous batching (CB) work well on Apple Silicon (MPS) with the metal-flash-sdpa kernel: 1. Enable paged decode fast path for hub kernels (initialization.py) ensure_decode_fast_path_is_available previously only allowed the paged decode fast path (block tables + flash" + }, + { + "n": 47779, + "d": "2026-08-31", + "t": "Refactor GGUF to speed-up inference", + "st": "OPEN", + "by": "SunMarc", + "a": 2435, + "r": 137, + "f": 35, + "c": 43, + "b": "This PR updates the GGUF in transformers in order to take advantages of some features that landed for the past few months and are now stable: conversion using WeightConverter, instead of our custom tensor processor. Support for loading quantized models. Before we were dequantizing by default. Add su" + } ], - "note": "GGUF side shipped; the MLX-kernels half of “MacOS first-class” is covered by transformers-to-mlx, not by in-repo kernels.", + "note": "Merged loading PRs plus four open kernel PRs — the bar runs to the latest commit on the open ones.", "links": [], - "d": null, + "d": "GGUF checkpoint loading shipped; the Metal/MPS kernel half is in review — optional Metal dequant fast path, MetalConfig for pre-quantized MLX checkpoints from the Hub, MPS paged-decode fast path for continuous batching, and a GGUF inference-speed refactor.", "span": [ "2025-06-06", - "2025-09-24" - ] + "2026-08-31" + ], + "roll": { + "owners": [ + [ + "ArthurZucker", + 2 + ], + [ + "44670", + 1 + ], + [ + "ctcanbol", + 1 + ], + [ + "HaroldBenoit", + 1 + ] + ], + "n": 8, + "add": 3145, + "del": 182, + "files": 54, + "commits": 74 + } }, { "t": "Quantization-aware training", "tag": "partial", "s": "wip", "prs": [ - [ - 41897, - "2025-11-04", - "[FPQuant] MXFP8 and MXFP4 backwards support", - "MERGED" - ] + { + "n": 41897, + "d": "2025-11-04", + "t": "[FPQuant] MXFP8 and MXFP4 backwards support", + "st": "MERGED", + "by": "BlackSamorez", + "a": 20, + "r": 5, + "f": 5, + "c": 6, + "b": "This PR adds MXFP4 and MXFP8 backwards support in combination with MXFP4 forward, allowing for lightning-fast QAT on Blackwell GPUs. It's blocked by QuTLASS v0.2.0 and fp_quant v0.3.0 release we're hoping to release within a day or two. Before submitting This PR fixes a typo or improves the docs (yo" + } ], "note": "MXFP8/MXFP4 backwards landed; QAT / QAD as scoped on the board did not.", "links": [], @@ -1360,7 +3246,20 @@ const TRACKS = [ "span": [ "2025-10-27", "2025-11-04" - ] + ], + "roll": { + "owners": [ + [ + "BlackSamorez", + 1 + ] + ], + "n": 1, + "add": 20, + "del": 5, + "files": 5, + "commits": 6 + } } ] }, @@ -1374,18 +3273,30 @@ const TRACKS = [ "tag": "CircleCI → GitHub Actions", "s": "done", "prs": [ - [ - 46721, - "2026-06-17", - "ci: disable CircleCI by replacing config with no-op", - "MERGED" - ], - [ - 48056, - "2026-08-18", - "[CircleCI] Enable CI for private forks, no-op for public repo", - "MERGED" - ] + { + "n": 46721, + "d": "2026-06-17", + "t": "ci: disable CircleCI by replacing config with no-op", + "st": "MERGED", + "by": "ydshieh", + "a": 241, + "r": 228, + "f": 3, + "c": 3, + "b": "Replaces the full CircleCI config with a minimal no-op job that immediately passes, and disables the CircleCI failure summary workflow. Why not just delete the config? Removing config.yml causes CircleCI to post a failing \"no configuration found\" check. Replacing it with a no-op keeps the check gree" + }, + { + "n": 48056, + "d": "2026-08-18", + "t": "[CircleCI] Enable CI for private forks, no-op for public repo", + "st": "MERGED", + "by": "ydshieh", + "a": 201, + "r": 14, + "f": 2, + "c": 5, + "b": "Summary Public repo (huggingface/transformers): genuine noop workflow — CircleCI does nothing (GitHub Actions handles CI here) Private fork (e.g. new model additions): full CI runs automatically, gated by [run-circleci] in the commit message to control costs Add peft_integration CircleCI job to matc" + } ], "note": "`.circleci/config.yml` still exists in the tree as a deliberate no-op.", "links": [], @@ -1393,37 +3304,74 @@ const TRACKS = [ "span": [ "2026-06-17", "2026-08-18" - ] + ], + "roll": { + "owners": [ + [ + "ydshieh", + 2 + ] + ], + "n": 2, + "add": 442, + "del": 242, + "files": 5, + "commits": 8 + } }, { "t": "Grafana CI dashboard in every PR", "tag": "dashboard link on each PR", "s": "done", "prs": [ - [ - 46410, - "2026-06-04", - "Add workflow to post CI Grafana dashboard link to PR", - "MERGED" - ], - [ - 46412, - "2026-06-04", - "Improve CI dashboard comment: rename and deduplicate", - "MERGED" - ], - [ - 46415, - "2026-06-04", - "ci: mention code quality failure in CI dashboard comment", - "MERGED" - ], - [ - 46774, - "2026-06-28", - "Insert a Grafana badge in the PR", - "MERGED" - ] + { + "n": 46410, + "d": "2026-06-04", + "t": "Add workflow to post CI Grafana dashboard link to PR", + "st": "MERGED", + "by": "ydshieh", + "a": 39, + "r": 0, + "f": 1, + "c": 1, + "b": "Summary Adds pr-ci-post-dashboard-link.yml workflow that triggers via workflow_run on completion of \"PR CI\" Uses actions/github-script to find the PR by head SHA and post a comment with the Grafana observability dashboard link Works for both same-repo and fork PRs since workflow_run always runs in t" + }, + { + "n": 46412, + "d": "2026-06-04", + "t": "Improve CI dashboard comment: rename and deduplicate", + "st": "MERGED", + "by": "ydshieh", + "a": 18, + "r": 1, + "f": 1, + "c": 1, + "b": "Summary Renames comment from **CI Observability Dashboard:** to **CI Dashboard:** Before posting, deletes any existing dashboard comment on the PR (handles both old and new text) to avoid duplicates on CI reruns" + }, + { + "n": 46415, + "d": "2026-06-04", + "t": "ci: mention code quality failure in CI dashboard comment", + "st": "MERGED", + "by": "ydshieh", + "a": 16, + "r": 1, + "f": 1, + "c": 1, + "b": "Summary When check_code_quality fails, the CI dashboard comment posted to the PR now includes a warning explaining that all test jobs were skipped The Grafana link is always posted regardless, so partial OTEL data is still accessible How it works After the \"PR CI\" workflow completes, listJobsForWork" + }, + { + "n": 46774, + "d": "2026-06-28", + "t": "Insert a Grafana badge in the PR", + "st": "MERGED", + "by": "tarekziade", + "a": 714, + "r": 73, + "f": 3, + "c": 4, + "b": "Moved the script to Python, in utils, and insert a badge in the PR" + } ], "note": null, "links": [ @@ -1436,25 +3384,54 @@ const TRACKS = [ "span": [ "2025-08-05", "2026-07-12" - ] + ], + "roll": { + "owners": [ + [ + "ydshieh", + 3 + ], + [ + "tarekziade", + 1 + ] + ], + "n": 4, + "add": 787, + "del": 75, + "files": 6, + "commits": 7 + } }, { "t": "mlinter findings as PR review comments", "tag": "inline review bot", "s": "done", "prs": [ - [ - 47819, - "2026-08-06", - "Post mlinter findings as inline PR review comments", - "MERGED" - ], - [ - 48117, - "2026-08-19", - "Enable mlinter findings artifact for inline PR reviews", - "MERGED" - ] + { + "n": 47819, + "d": "2026-08-06", + "t": "Post mlinter findings as inline PR review comments", + "st": "MERGED", + "by": "ydshieh", + "a": 198, + "r": 6, + "f": 2, + "c": 2, + "b": "What this does Adds a post-mlinter-review job to pr-ci-post-dashboard-link.yml that: Downloads the mlinter-findings artifact produced by the check_code_quality job Posts the findings as a single PR review with inline comments anchored to the diff The review includes a summary table (rule → count) an" + }, + { + "n": 48117, + "d": "2026-08-19", + "t": "Enable mlinter findings artifact for inline PR reviews", + "st": "MERGED", + "by": "ydshieh", + "a": 1, + "r": 1, + "f": 1, + "c": 1, + "b": "Summary Last piece of the mlinter inline review pipeline, verified in #48109. Adds --output-json mlinter-findings.json to the mlinter invocation so the check_code_quality job uploads a mlinter-findings artifact. The post-mlinter-review job (already on main) then downloads it and posts inline review " + } ], "note": null, "links": [], @@ -1462,31 +3439,62 @@ const TRACKS = [ "span": [ "2026-08-06", "2026-08-19" - ] + ], + "roll": { + "owners": [ + [ + "ydshieh", + 2 + ] + ], + "n": 2, + "add": 199, + "del": 7, + "files": 3, + "commits": 3 + } }, { "t": "Benchmarks + perf regression CI", "tag": "regression detection", "s": "done", "prs": [ - [ - 39198, - "2025-07-16", - "CI workflow for performed test regressions", - "MERGED" - ], - [ - 39768, - "2025-08-15", - "Benchmarking improvements", - "MERGED" - ], - [ - 41916, - "2025-11-07", - "feat(ci): add continuous batching to benchmarks", - "MERGED" - ] + { + "n": 39198, + "d": "2025-07-16", + "t": "CI workflow for performed test regressions", + "st": "MERGED", + "by": "ahadnagy", + "a": 183, + "r": 13, + "f": 3, + "c": 5, + "b": "This PR adds a new workflow to the CI that dumps a report about the differences in performed tests. We noticed that the test counts sometimes fluctuate quite a bit between CI runs, and this report will help in narrowing down the root cause and find regressions (removed test, catastrophic failure dur" + }, + { + "n": 39768, + "d": "2025-08-15", + "t": "Benchmarking improvements", + "st": "MERGED", + "by": "ahadnagy", + "a": 719, + "r": 435, + "f": 7, + "c": 7, + "b": "The goal of this PR is to start improving the benchmarking infrastructure of Transformers. In the short-term we'd like to build a dataset similar to what Diffusers did (https://huggingface.co/datasets/diffusers/benchmarks) and put a dashboard on Spaces. This PR starts with the data acquisition part " + }, + { + "n": 41916, + "d": "2025-11-07", + "t": "feat(ci): add continuous batching to benchmarks", + "st": "MERGED", + "by": "McPatate", + "a": 190, + "r": 107, + "f": 10, + "c": 15, + "b": "" + } ], "note": null, "links": [], @@ -1494,19 +3502,42 @@ const TRACKS = [ "span": [ "2025-07-03", "2025-11-07" - ] + ], + "roll": { + "owners": [ + [ + "ahadnagy", + 2 + ], + [ + "McPatate", + 1 + ] + ], + "n": 3, + "add": 1092, + "del": 555, + "files": 20, + "commits": 27 + } }, { "t": "Tiny-model integration testing", "tag": "cheap end-to-end tests", "s": "wip", "prs": [ - [ - 47005, - "2026-07-03", - "Add tiny_model_id support to ProcessorTesterMixin for memory-sensitive tests", - "MERGED" - ] + { + "n": 47005, + "d": "2026-07-03", + "t": "Add tiny_model_id support to ProcessorTesterMixin for memory-sensitive tests", + "st": "MERGED", + "by": "ydshieh", + "a": 82, + "r": 30, + "f": 3, + "c": 4, + "b": "Summary Adds tiny_model_id attribute to ProcessorTesterMixin allowing test classes to specify a lightweight Hub repo (e.g. containing only a tiny tokenizer) for use in most tests, while keeping model_id as the full Hub repo used only when explicitly requested via get_processor(use_full=True) / get_c" + } ], "note": null, "links": [], @@ -1514,31 +3545,62 @@ const TRACKS = [ "span": [ "2026-07-01", "2026-07-03" - ] + ], + "roll": { + "owners": [ + [ + "ydshieh", + 1 + ] + ], + "n": 1, + "add": 82, + "del": 30, + "files": 3, + "commits": 4 + } }, { "t": "timm wrapper", "tag": "timm models as backbones", "s": "done", "prs": [ - [ - 34729, - "2024-11-19", - "feat: allow to use hf-models", - "MERGED" - ], - [ - 34564, - "2024-12-11", - "Add TimmWrapper", - "MERGED" - ], - [ - 35794, - "2025-01-21", - "Auto-add `timm` tag to timm-wrapper models.", - "MERGED" - ] + { + "n": 34729, + "d": "2024-11-19", + "t": "feat: allow to use hf-models", + "st": "MERGED", + "by": "cgebbe", + "a": 0, + "r": 5, + "f": 1, + "c": 1, + "b": "Problem: This doesn't currently work, but should in my opinion? import transformers model = transformers.AutoBackbone.from_pretrained( \"hf-hub:bioptimus/H-optimus-0\", use_timm_backbone=True, use_pretrained_backbone=True, ) Solution approach: Just remove the two lines which explicitly check whether t" + }, + { + "n": 34564, + "d": "2024-12-11", + "t": "Add TimmWrapper", + "st": "MERGED", + "by": "qubvel", + "a": 1432, + "r": 129, + "f": 25, + "c": 92, + "b": "Adds a TimmWrapper set of classes such that timm models can be loaded in as transformer models into the library. Continue of #33687 General Usage import torch from urllib.request import urlopen from PIL import Image from transformers import AutoConfig, AutoModelForImageClassification, AutoImageProce" + }, + { + "n": 35794, + "d": "2025-01-21", + "t": "Auto-add `timm` tag to timm-wrapper models.", + "st": "MERGED", + "by": "pcuenca", + "a": 1, + "r": 0, + "f": 1, + "c": 1, + "b": "Works for fine-tuned or exported models: from transformers import AutoModelForImageClassification checkpoint = \"timm/vit_base_patch16_224.augreg2_in21k_ft_in1k\" model = AutoModelForImageClassification.from_pretrained(checkpoint) model.push_to_hub(\"pcuenq/tw1\") The uploaded model will now show snippe" + } ], "note": null, "links": [], @@ -1546,7 +3608,28 @@ const TRACKS = [ "span": [ "2024-11-01", "2025-01-21" - ] + ], + "roll": { + "owners": [ + [ + "qubvel", + 1 + ], + [ + "cgebbe", + 1 + ], + [ + "pcuenca", + 1 + ] + ], + "n": 3, + "add": 1433, + "del": 134, + "files": 27, + "commits": 94 + } }, { "t": "hf jobs / closer hub CI integration", @@ -1556,7 +3639,15 @@ const TRACKS = [ "note": "No PRs found in transformers as of Sep 2026.", "links": [], "d": null, - "span": null + "span": null, + "roll": { + "owners": [], + "n": 0, + "add": 0, + "del": 0, + "files": 0, + "commits": 0 + } } ] }, @@ -1570,18 +3661,30 @@ const TRACKS = [ "tag": "torch-native preprocessing", "s": "done", "prs": [ - [ - 45795, - "2026-05-18", - "[docs] adding audio/video processors", - "MERGED" - ], - [ - 46500, - "2026-06-08", - "Fix `{image/video/audio}_token_ids` in `ProcessorMixin`", - "MERGED" - ] + { + "n": 45795, + "d": "2026-05-18", + "t": "[docs] adding audio/video processors", + "st": "MERGED", + "by": "stevhliu", + "a": 255, + "r": 9, + "f": 3, + "c": 4, + "b": "follow up to this comment about separate docs for how to add audio and video processors" + }, + { + "n": 46500, + "d": "2026-06-08", + "t": "Fix `{image/video/audio}_token_ids` in `ProcessorMixin`", + "st": "MERGED", + "by": "hmellor", + "a": 18, + "r": 0, + "f": 1, + "c": 1, + "b": "#45493 converted these attributes to properties. This breaks custom models which set {image/video/audio}_token_ids directly because: There is no setter, so these properties are read-only The getter will only ever read from image_token_id This PR maintains BC by: Adding a setter for these properties " + } ], "note": null, "links": [], @@ -1589,19 +3692,42 @@ const TRACKS = [ "span": [ "2026-05-05", "2026-06-08" - ] + ], + "roll": { + "owners": [ + [ + "stevhliu", + 1 + ], + [ + "hmellor", + 1 + ] + ], + "n": 2, + "add": 273, + "del": 9, + "files": 4, + "commits": 5 + } }, { "t": "Native mamba & hybrid cache", "tag": "SSM + attention hybrids", "s": "done", "prs": [ - [ - 44950, - "2026-03-31", - "🚨 [Cache] Native mamba & hybrid cache", - "MERGED" - ] + { + "n": 44950, + "d": "2026-03-31", + "t": "🚨 [Cache] Native mamba & hybrid cache", + "st": "MERGED", + "by": "Cyrilvallez", + "a": 1346, + "r": 4113, + "f": 64, + "c": 59, + "b": "As per the title. This PR finally makes mamba layer caches first class citizen, and adds native support for them. It supports the following layers combinations: all mamba layers alternating attention layer/mamba layer layers that are BOTH mamba and attention (zamba models) For this, it adds the 2 fo" + } ], "note": null, "links": [], @@ -1609,7 +3735,20 @@ const TRACKS = [ "span": [ "2026-03-23", "2026-03-31" - ] + ], + "roll": { + "owners": [ + [ + "Cyrilvallez", + 1 + ] + ], + "n": 1, + "add": 1346, + "del": 4113, + "files": 64, + "commits": 59 + } }, { "t": "Full vision + multimodal fine-tuning", @@ -1619,7 +3758,15 @@ const TRACKS = [ "note": "No dedicated PRs found; tracked on the board for H2 2026.", "links": [], "d": null, - "span": null + "span": null, + "roll": { + "owners": [], + "n": 0, + "add": 0, + "del": 0, + "files": 0, + "commits": 0 + } } ] }