Commit Graph

  • 93f4d08b90 try build tiny ydshieh 2026-02-22 20:59:58 +08:00
  • 931616922a try build tiny ydshieh 2026-02-22 19:43:51 +08:00
  • 1d798f77c5 try build tiny ydshieh 2026-02-22 18:26:41 +08:00
  • ccf11a6dc1 try build tiny ydshieh 2026-02-22 09:39:09 +08:00
  • b14f008810 try build tiny ydshieh 2026-02-22 09:05:18 +08:00
  • f767b4e78e try build tiny ydshieh 2026-02-21 22:55:10 +08:00
  • 060c3e02d1 try build tiny ydshieh 2026-02-21 08:51:57 +08:00
  • c860ada589 try build tiny ydshieh 2026-02-21 08:16:16 +08:00
  • 343651c903 try build tiny ydshieh 2026-02-20 23:03:51 +08:00
  • ff2c7b43c0 try build tiny ydshieh 2026-02-20 22:22:30 +08:00
  • 8f6c2d9a8f try build tiny ydshieh 2026-02-20 21:10:13 +08:00
  • 894bc63c64 try build tiny ydshieh 2026-02-20 21:00:28 +08:00
  • c91a72acbf auto modular detection + conversion + pr ita.zaporozhets@huggingface.co 2026-03-27 23:47:42 +00:00
  • 0a2c4f9924 fix copies Arthur 2026-03-28 00:37:28 +01:00
  • 8d445e599a fix? Arthur 2026-03-28 00:33:52 +01:00
  • 5df7f825f6 converter up Arthur 2026-03-27 21:29:56 +01:00
  • 5ef4900f98 arf Arthur 2026-03-27 21:29:50 +01:00
  • f5956fded0 fix check copies Arthur 2026-03-27 19:33:36 +01:00
  • 58ddf7f07b style Arthur 2026-03-27 18:12:06 +01:00
  • 9a9997fd73 Remove unused TensorFlow env var (#45065) Sai-Suraj-27 2026-03-27 22:41:54 +05:30
  • b7473083b0 fix imports Arthur 2026-03-27 18:11:28 +01:00
  • ecdf95c66c fix: add identity reverse_op to dequantize ops for save_pretrained (#44983) Hyungkeun Park 2026-03-28 02:09:03 +09:00
  • 1997c0eef4 lets used isolated envs tarekziade-optimize-imports Tarek Ziade 2026-03-27 18:05:32 +01:00
  • a5c748104d update Arthur 2026-03-27 17:54:13 +01:00
  • 2ef9b593ce shard on imports too Tarek Ziade 2026-03-27 17:43:57 +01:00
  • 9b4cf6af5e adding timers Tarek Ziade 2026-03-27 17:38:58 +01:00
  • 78e8741271 make it idempotent Tarek Ziade 2026-03-27 17:32:20 +01:00
  • b514e5d656 put all import checks in the same spot Tarek Ziade 2026-03-27 17:30:21 +01:00
  • 38d50213c0 shard checkers Tarek Ziade 2026-03-27 17:25:45 +01:00
  • 12b6b57cac Fix when RoPE params are in kwargs (#45049) Raushan Turganbay 2026-03-27 17:15:27 +01:00
  • b15369eaf8 up Arthur 2026-03-27 17:11:05 +01:00
  • e8ea722d9c fix repo Arthur 2026-03-27 16:57:45 +01:00
  • 5a0e02d7aa more Arthur 2026-03-27 16:56:16 +01:00
  • 678dbc56e8 up Arthur 2026-03-27 16:53:54 +01:00
  • d65c2b138a chore: update update_metdata.yml (#45054) hf-security-analysis[bot] 2026-03-27 16:51:05 +01:00
  • ca6acc7849 [FA] Fix BC support for a few versions + add deprecation cycle (#45061) Anton Vlasjuk 2026-03-27 16:30:33 +01:00
  • f9d73c9740 fixes Arthur 2026-03-27 16:23:58 +01:00
  • b0bba2d832 fix(testing): Fix Parakeet, Evolla, Pi0, and Phi-3 test failures on main CI (#45004) Harshal Janjani 2026-03-27 19:06:08 +04:00
  • 2f0267a351 style Arthur 2026-03-27 16:05:49 +01:00
  • 756aa7ce44 update Arthur 2026-03-27 16:02:40 +01:00
  • 44686173b2 Allow advanced users to override model_type in AutoConfig.from_pretrained (#45058) Harry Mellor 2026-03-27 14:17:31 +00:00
  • 08062754e2 styling fix-auto-doc Arthur 2026-03-27 15:15:01 +01:00
  • 13f5646527 fix Arthur 2026-03-27 15:10:48 +01:00
  • ce4a791c52 Fix llama4 bnb mode (#44588) jiqing-feng 2026-03-27 22:05:40 +08:00
  • cc4ef19bb8 Fix failing SmolLM3IntegrationTest (#45048) Sai-Suraj-27 2026-03-27 19:33:26 +05:30
  • 9cd278715c fix tests/quantization/fp_quant_integration/test_fp_quant.py::FPQuant… (#44644) Wang, Yi 2026-03-27 21:56:47 +08:00
  • d266fedc3e fix-repo Arthur 2026-03-27 12:58:18 +01:00
  • 6e906b5491 fiixup Arthur 2026-03-27 12:56:27 +01:00
  • a1bda86f66 more import cleanup Arthur 2026-03-27 12:55:56 +01:00
  • 4803b722e2 long due Arthur 2026-03-27 12:35:20 +01:00
  • 6e011bf71a beit is torchvisionbackend Arthur 2026-03-27 12:34:31 +01:00
  • 4099913502 marvellous that's how we protect torch :) Arthur 2026-03-27 12:30:33 +01:00
  • c078455916 fix import shinanigans Arthur 2026-03-27 12:29:10 +01:00
  • ed3111a6db remove if typechecking Arthur 2026-03-27 12:19:24 +01:00
  • be82ad82ba remove torch when its not necessary Arthur 2026-03-27 12:16:40 +01:00
  • 689f52ce6b chore: remove old extras (#45024) Tarek Ziade 2026-03-27 12:16:01 +01:00
  • 1417bc1917 update Arthur 2026-03-27 12:14:11 +01:00
  • 1cb0950529 Merge branch 'main' into cb-batched-logits Rémi Ouazan 2026-03-27 12:12:20 +01:00
  • 0efcf1b069 Avoid Image.open failure (#44645) Wang, Yi 2026-03-27 19:11:05 +08:00
  • 035abaa1d4 [Bugfix] Remove redundant @requires(backends=("vision",)) from PIL backends Lidang-Jiang 2026-03-27 18:24:59 +08:00
  • cf6028853b replaced the imported-module analysis path in convert_modular_file() with a metadata-free mapper visit that tracks top-level context and source order directly, eliminating expensive LibCST ParentNodeProvider/PositionProvider passes while preserving byte-identical output tarekziade-modular-speedup Tarek Ziade 2026-03-27 11:22:12 +01:00
  • 0573124fc1 chore: Fix mlinter cache location (#45052) Tarek Ziade 2026-03-27 11:17:02 +01:00
  • 07f2a0ad5a Python-only fast path based on symtable plus a lightweight AST pass Tarek Ziade 2026-03-27 10:44:26 +01:00
  • 4ee7f51e46 Embedding VLMs don't need a head (#45000) Raushan Turganbay 2026-03-27 10:43:28 +01:00
  • ba378c2c7c Style and doc remi-or 2026-03-27 09:41:39 +00:00
  • 05514c4bb6 Fix GraniteConfig type hints to accept int for multiplier fields (#45019) Javier De Jesus 2026-03-27 10:20:14 +01:00
  • 46d70f4145 second optim / cache Tarek Ziade 2026-03-27 10:15:26 +01:00
  • 7b00e3ba39 fix: preserve rotary_pct across save/load cycle in GPTNeoX configs (#44985) Krishna Chaitanya 2026-03-27 02:05:08 -07:00
  • faf2098288 investigate modular conversion speedups Tarek Ziade 2026-03-27 09:31:19 +01:00
  • 1abefc9cd4 [Bugfix] Remove incorrect torchvision requirement from PIL backend image processors Lidang-Jiang 2026-03-27 16:17:09 +08:00
  • 23773e7140 refactor: speed up docstring checker (#45009) Tarek Ziade 2026-03-27 08:13:17 +01:00
  • 1f553bdc17 post release tag update Arthur 2026-03-27 01:25:24 +01:00
  • 276f140202 v5.4.0 v5.4.0 v5.4-release Arthur 2026-03-26 18:37:09 +01:00
  • 435203ec55 style was missing sorry @ydshieh :) (#45038) Arthur 2026-03-27 01:19:02 +01:00
  • 16d437e43a fix: protect torch imports in processing files and fix import guards Arthur 2026-03-27 01:14:51 +01:00
  • 97b7727e11 Fix release full (#45029) Arthur 2026-03-27 01:11:22 +01:00
  • 78bdaf0b39 Add cohere asr (#45023) eustlb 2026-03-26 23:48:11 +01:00
  • fe01b1cd59 add kwargs remi-or 2026-03-26 22:18:41 +00:00
  • 2fa396db08 Review remi-or 2026-03-26 16:55:25 +00:00
  • 8cc5db1d4a Fix final tests remi-or 2026-03-26 16:33:35 +00:00
  • 64ac0e2aa3 Fix tests remi-or 2026-03-26 14:14:04 +00:00
  • 4cfd983ffe fix for multiple processors remi-or 2026-03-26 09:01:04 +00:00
  • 08e8108605 fix for temperature remi-or 2026-03-25 08:59:35 +00:00
  • 487922a224 nit remi-or 2026-03-25 08:00:32 +00:00
  • 9028936e7c Fix test remi-or 2026-03-24 17:39:22 +00:00
  • aca477aaf5 omw to fix remi-or 2026-03-24 16:16:23 +00:00
  • 157cad54a2 Fix metrics remi-or 2026-03-24 15:10:31 +00:00
  • ed49ac9508 Update logprobs test remi-or 2026-03-24 15:10:21 +00:00
  • c74fef1248 Stacked and rebased remi-or 2026-03-26 22:23:30 +00:00
  • e895107784 docs: add PermuteForRope to conversion operations reverse table (#45035) Surya Teja Addanki 2026-03-26 16:54:05 -05:00
  • 67100ccaad [CB] Persistent manager (#44435) Rémi Ouazan 2026-03-26 22:47:57 +01:00
  • 69f9d552e9 Add BC for _further_process_kwargs (#45033) Harry Mellor 2026-03-26 20:48:50 +00:00
  • 4fdd1255ff check check_workflow ydshieh 2026-03-26 20:02:56 +01:00
  • 19b2f0230b check ydshieh 2026-03-26 20:01:37 +01:00
  • e3f7cc3b4a Use multi runners to check new failing tests in a CI run (#45032) Yih-Dar 2026-03-26 19:58:10 +01:00
  • 882ffdbbd6 [fix] Use the correct _tied_weights_keys for CamembertForCausalLM (#45031) Tom Aarsen 2026-03-26 19:50:02 +01:00
  • 9b6695a44f Merge fsdp-core-model-loading into refactor-tp-dtensor refactor-tp-dtensor 3outeille 2026-03-26 17:48:41 +00:00
  • 607cc11491 DTensor-based TP + FSDP2 shard-on-read composability fsdp-core-model-loading 3outeille 2026-03-26 17:47:52 +00:00
  • d81ad48109 change dev ver. we forgot to do this when we released 5.3.0 Arthur 2026-03-26 18:35:33 +01:00
  • d48fcc7eee FSDP + TP now works but that requires us to rely on dtensor TP 3outeille 2026-03-26 16:48:04 +00:00