Commit Graph

  • 28d3bd618c Merge branch 'main' into fix_qwen3_omni_config Anton Vlasjuk 2026-03-25 15:18:42 +01:00
  • 38605deacc Add sarvam model (auto-generated modular integration) add-sarvam-model ita.zaporozhets@huggingface.co 2026-03-25 13:58:57 +00:00
  • 2f62491792 fix: remove Copied from comments between @torch.jit.script and def for Python 3.13 compat (#44986) Krishna Chaitanya 2026-03-25 06:31:30 -07:00
  • 0e1978c9eb More small vllm fixes (#44990) Arthur 2026-03-25 14:05:41 +01:00
  • b9a547d37a fix get_rope_index models and chunked vision attn models IlyasMoutawwakil 2026-03-25 13:21:18 +01:00
  • 2620c4ddd4 fix(models): Fix Perceiver interpolate_pos_encoding interpolating to the source size (#44899) Harshal Janjani 2026-03-25 15:41:05 +04:00
  • c532659b87 Allow mm_token_type be non-padded lists (#44563) Raushan Turganbay 2026-03-25 12:20:06 +01:00
  • 7c843391b7 init 3outeille 2026-03-25 09:09:34 +00:00
  • c3b41c6c57 Fix CPU 16 bytes alignment issue using equivalent fallback (#44970) Ilyas Moutawwakil 2026-03-25 09:45:47 +01:00
  • b3d7942fba refactor: unify QA calls (#44879) Tarek Ziade 2026-03-25 09:39:39 +01:00
  • 1ee312adb2 push to docker again [push-ci-image] push-ci-image Tarek Ziade 2026-03-25 09:30:21 +01:00
  • 28af8184fb Fix tie_word_embedding issues with Qwen2VL (#44976) Harry Mellor 2026-03-24 20:48:14 +00:00
  • 28e1cc59ec Support Modular (!!) + Configs in check_auto_docstrings (#44803) Yoni Gozlan 2026-03-24 13:46:05 -04:00
  • 692d187dc7 [ vllm x v5] nit (#44971) Arthur 2026-03-24 18:40:02 +01:00
  • 18f88de7f5 LwDetrImageLoss: Fix dtype casting to prevent crash when using amp on cuda device (#44886) Matt 2026-03-24 17:47:07 +01:00
  • b94fee0493 [AMD CI] Gemma3/Gemma3n Expectations (#44972) BADAOUI Abdennacer 2026-03-24 17:20:55 +01:00
  • 15b3b8924b Officially launch parse_response (#44674) Matt 2026-03-24 15:36:25 +00:00
  • e22f3088dd fix load_best_model_checkpoint_at_end do not load the best model chec… (#44583) wilnn 2026-03-24 11:28:32 -04:00
  • 82106705eb fix: split MXFP4 dependency checks for specific error messages (#44930) Javier De Jesus 2026-03-24 16:17:30 +01:00
  • 2d63f8897d Add modular_sarvam.py skeleton based on deepseek_v3 add-modular-sarvam ita.zaporozhets@huggingface.co 2026-03-24 14:43:30 +00:00
  • a6bfaeee3f push to docker [push-ci-image] Tarek Ziade 2026-03-24 15:41:17 +01:00
  • f88340152e temp revert dequant-fix Arthur 2026-03-24 15:39:44 +01:00
  • a8732d5546 Fix failing T5ModelIntegrationTest (#44934) Sai-Suraj-27 2026-03-24 20:06:58 +05:30
  • d4895f0810 Config kwargs (#44953) Raushan Turganbay 2026-03-24 15:14:45 +01:00
  • 44b5b6718b added Make to the docker image and tomli to the quality extra [build-ci-image] Tarek Ziade 2026-03-24 12:41:33 +01:00
  • efa88eab1b Remove auto from bases in modular eval dataset ita.zaporozhets@huggingface.co 2026-03-24 13:09:24 +00:00
  • 2f121eaf3c Fix: Update optimization.py (#44909) anshuS1310 2026-03-24 18:22:46 +05:30
  • 07ddf7b79d update index database itazap 2026-03-24 12:37:18 +01:00
  • 4f8de83424 [CB] [Minor] Simplify test suite (#44858) Rémi Ouazan 2026-03-24 12:32:00 +01:00
  • fc02c66387 try test_circleci_api ydshieh 2026-03-24 12:19:16 +01:00
  • 53a0404ac6 try ydshieh 2026-03-24 12:12:54 +01:00
  • 4fb2247cc9 try ydshieh 2026-03-24 12:07:55 +01:00
  • cb5a836ca1 try ydshieh 2026-03-24 12:00:36 +01:00
  • 9ef15d7016 Allow arbitrary template kwargs in processors (#44881) Raushan Turganbay 2026-03-24 11:59:58 +01:00
  • 8610e8b733 try ydshieh 2026-03-24 11:58:51 +01:00
  • 6a056a16a8 Fix missing post_processor in DebertaV2Tokenizer causing no special t… (#44570) Kirill Ustinov 2026-03-24 12:31:46 +03:00
  • bc8c80e028 incorrect model list update (#44880) Ita Zaporozhets 2026-03-24 10:15:35 +01:00
  • a48a63c27d refactor: mlinter as its own package (#44939) Tarek Ziade 2026-03-24 08:47:30 +01:00
  • 19bba39fa7 [CB] Add an option to return logprobs (#44835) Rémi Ouazan 2026-03-23 19:21:05 +01:00
  • a9532bcf88 [docs] peft (#44804) Steven Liu 2026-03-23 10:05:59 -07:00
  • dda546845f Continuous batching thread safety (#44924) Qubitium-ModelCloud 2026-03-24 00:13:24 +08:00
  • e5ad394620 Add static FP8 expert support (#44895) Marc Sun 2026-03-23 15:50:55 +01:00
  • f6195948b4 Fix variable shadowing in pipeline example and typo in BART docs (BERT → BART) (#44935) Vanshika Sohal 2026-03-23 19:48:54 +05:30
  • 0f19dec775 Fix failing job Update Transformers metadata after #43514 (#44941) Yih-Dar 2026-03-23 14:41:37 +01:00
  • ed003b4482 Clearer type hints and fix rope validation in configs (#44943) Raushan Turganbay 2026-03-23 14:20:04 +01:00
  • eb3d67aaaf Correct docstrings for from_pretrained (url input deprecated) (#44946) Bart Schilperoort 2026-03-23 13:50:29 +01:00
  • ddb709c8c9 fix(i18n): replace broken relative links to awesome-transformers.md with absolute URLs (#44905) Jianwei Zhang 2026-03-23 20:35:33 +08:00
  • d03945fb8d Fix backward compatibility for full path imports of Fast Image Processors (#44926) Yoni Gozlan 2026-03-23 08:22:55 -04:00
  • 29db503cde chore(typing): added rule 11 (#44865) Tarek Ziade 2026-03-23 13:19:45 +01:00
  • c7199451a8 all tests passing eustlb 2026-03-23 12:18:38 +01:00
  • b1527a32a1 fix: improve processor loading performance by avoiding redundant tokenizer parsing (#44927) Yih-Dar 2026-03-23 11:46:49 +01:00
  • 9dc8d8aa30 fix(camembert): add tie_word_embeddings=True to CamembertConfig (#44931) r266-tech 2026-03-23 18:47:48 +08:00
  • 687a70daa9 Support SizeDict import in get_size_dict (#44903) Yoni Gozlan 2026-03-23 06:16:30 -04:00
  • a868375665 fix processing_utils.py: avoid deepcopying tokenizer in ProcessorMixin to improve performance (#44894) Yih-Dar 2026-03-23 10:56:33 +01:00
  • 55cc1a7fb8 fix: set clean_up_tokenization_spaces=False in Llama 3 tokenizer conversion (#44914) maxsloef-goodfire 2026-03-23 01:28:31 -07:00
  • 3502ae4ac3 style Arthur 2026-03-23 08:29:57 +01:00
  • 15fbdacf4e Merge branch 'main' of github.com:huggingface/transformers into dequant-fix Arthur 2026-03-23 08:07:09 +01:00
  • 394b1a0eaa Add big angry code agent warnings! (#44890) Matt 2026-03-22 18:48:30 +00:00
  • a04a9b98c1 Fix no_split_modules Sai-Suraj-27 2026-03-22 10:40:49 +00:00
  • d7a6fb3b3a Merge branch 'main' of github.com:huggingface/transformers into fix_qwen3_omni_config Sai-Suraj-27 2026-03-22 10:26:02 +00:00
  • 3a3b59cb1a [docs] model cards (#44837) Steven Liu 2026-03-20 15:33:10 -07:00
  • 52bc9b7e4d [Model] Add UVDoc Model Support (#43385) XingweiDeng 2026-03-21 06:02:35 +08:00
  • e168f86efb Add backward compatibility for direct imports from legacy image_processing_utils_fast (#44897) Yoni Gozlan 2026-03-20 15:47:10 -04:00
  • e6ed96c7e9 [FA4] Add kernels fallback (#44797) Anton Vlasjuk 2026-03-20 19:50:51 +01:00
  • a269c990e8 Bump kernels version dependency to avoid crashes (#44887) Cyril Vallez 2026-03-20 19:49:41 +01:00
  • 38192b7c67 Merge remote-tracking branch 'origin/main' into fp8-per-tensor fp8-per-tensor Marc Sun 2026-03-20 18:10:13 +00:00
  • 83dc7d07e5 revert Marc Sun 2026-03-20 18:08:31 +00:00
  • a9bed19684 Merge branch 'main' into hf-exporters Ilyas Moutawwakil 2026-03-20 18:23:58 +01:00
  • d4f88c2062 [Model] Add SLANeXt Model Support (#43707) Liu Jiaxuan 2026-03-21 01:11:02 +08:00
  • 3677080924 simplify Marc Sun 2026-03-20 15:59:44 +00:00
  • fb72fd4e67 remove stream stream-nemotron Cyril Vallez 2026-03-20 15:44:23 +01:00
  • e87125508b Fix core dumped when NemotronH is torch compiled (#44854) Yih-Dar 2026-03-20 15:29:13 +01:00
  • 20a233bdc5 Fix several based models' pipeline parallel support (#44699) Harry Mellor 2026-03-20 13:43:07 +00:00
  • 81128509a5 fix(testing): Fix PaliGemma 2 and PaddleOCR-VL test failures on main (#44765) Harshal Janjani 2026-03-20 17:33:51 +04:00
  • 82db888e29 Fix dtype guessing from state dict (#44883) Cyril Vallez 2026-03-20 14:12:31 +01:00
  • 7cd9b985e0 Add missing dunder methods to SizeDict (#44884) Harry Mellor 2026-03-20 12:05:42 +00:00
  • 7c87d35371 fix in onnx exporter order and executorch backend preparation IlyasMoutawwakil 2026-03-20 12:56:05 +01:00
  • 4eee818214 revert patchtsmixer IlyasMoutawwakil 2026-03-20 12:55:24 +01:00
  • b7164eca86 Fix VL model rope_deltas batch size mismatch in online RL training (#44873) Sergio Paniego Blanco 2026-03-20 12:51:22 +01:00
  • 1229e90d46 Fix layer_types type hint for AFMoE and Llama4 (#44874) Harry Mellor 2026-03-20 11:50:31 +00:00
  • cef2830048 Align lfm2 cache to other mamba caches (#44866) llama4_config Cyril Vallez 2026-03-20 11:50:26 +01:00
  • 24d4735f67 better leaf tensors handling IlyasMoutawwakil 2026-03-20 11:28:34 +01:00
  • 0886f388ff revert deepseek vl IlyasMoutawwakil 2026-03-20 11:17:40 +01:00
  • 59044fd860 revert idefics IlyasMoutawwakil 2026-03-20 11:15:41 +01:00
  • 47fb0a5ade revert idefics IlyasMoutawwakil 2026-03-20 11:09:46 +01:00
  • f5e573080a Fix nemotron config docstrings (#44878) Cyril Vallez 2026-03-20 11:02:26 +01:00
  • 8dc7a52d76 Fix nemotron_h modular (#44876) Cyril Vallez 2026-03-20 11:00:32 +01:00
  • 4c94907265 fix lfm cache IlyasMoutawwakil 2026-03-20 10:49:10 +01:00
  • 2cd52c267c feat: added cache to the model linter (#44790) Tarek Ziade 2026-03-20 10:28:36 +01:00
  • 4d64807dad add back this comment tarekziade-cli-refacto Tarek Ziade 2026-03-20 10:16:37 +01:00
  • 24e303fd87 generated a module docstring to explain what the module does, it's hard to follow and complex Tarek Ziade 2026-03-20 10:11:16 +01:00
  • 67d335cdce keep the old name for backward compat even if it's unused anywhere in our codebase Tarek Ziade 2026-03-20 10:05:33 +01:00
  • bd0b91af62 refactor: in cli/server extract helpers, flatten nested functions, use CB context manager Tarek Ziade 2026-03-20 09:56:54 +01:00
  • 85de2da8b9 fix style IlyasMoutawwakil 2026-03-20 09:42:21 +01:00
  • 67a4d30f55 style IlyasMoutawwakil 2026-03-20 09:37:29 +01:00
  • aa1c36f1a9 [Model] Add PP-Chart2Table Model Support (#43767) XingweiDeng 2026-03-20 03:00:37 +08:00
  • b96f8a9896 [Mistral] Fix query scaling for Mistral4 and Ministral3 (#44860) Cyril Vallez 2026-03-19 19:02:04 +01:00
  • e94695e574 Propagate the model loading from transformers serve to chat (#44758) Lysandre Debut 2026-03-20 02:08:03 +09:00
  • b79d9d4988 fix Marc Sun 2026-03-19 16:44:49 +00:00
  • 03729cb627 style Marc Sun 2026-03-19 16:25:54 +00:00