Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Dec 19, 2025

• Fix tests trainer again (#42933)

  SunMarc

  authored
  7017994

  Copy full SHA for 7017994

• [`Ernie 4.5`] Ernie VL models (#39585)

  Show description for a8a2262
  

  vasqu

  and

  molbap

  authored
  a8a2262

  Copy full SHA for a8a2262

• [`Generation`] Fix default overwrite for non-`None` defaults (#42958)

  Show description for f218ed2
  vasqu

  authored
  f218ed2

  Copy full SHA for f218ed2

• Fix infinity in JSON serialized files (#42959)

  Show description for d14d99e
  LysandreJik

  authored
  d14d99e

  Copy full SHA for d14d99e

• Add runner specification to CodeQL workflow (#42955)

  paulinebm

  authored
  d54d78f

  Copy full SHA for d54d78f

• [loading] Really initialize on meta device for huge perf gains (#42941)

  Show description for bb9357f
  Cyrilvallez

  authored
  bb9357f

  Copy full SHA for bb9357f

• fix error: 'BlockMask' object has no attribute 'dtype' for lasr model (#42802)

  Show description for 789226c
  

  kaixuanliu

  and

  vasqu

  authored
  789226c

  Copy full SHA for 789226c

• fix concat order (#42946)

  eustlb

  authored
  a4d6229

  Copy full SHA for a4d6229

• [kernels] Fix failling tests (#42953)

  Show description for cf0f071
  MekkCyber

  authored
  cf0f071

  Copy full SHA for cf0f071

• [Quantization] CI green by end of year (#42951)

  Show description for b5eea34
  MekkCyber

  authored
  b5eea34

  Copy full SHA for b5eea34

Commits on Dec 18, 2025

• Updated `backbone_config` docstrings and type annotations (#42927)

  Show description for 3e4baf8
  simonreise

  authored
  3e4baf8

  Copy full SHA for 3e4baf8

• [docs] dtype (#42883)

  Show description for b93f2e3
  stevhliu

  authored
  b93f2e3

  Copy full SHA for b93f2e3

• Load generation config from nested configs (#42922)

  Show description for f2c6d2a
  zucchini-nlp

  authored
  f2c6d2a

  Copy full SHA for f2c6d2a

• 🚨 Fix ConvNeXt image processor default interpolation to BICUBIC (#42934)

  Show description for 60634ca
  

  lukepayyapilli

  and

  yonigozlan

  authored
  60634ca

  Copy full SHA for 60634ca

• fix(tvp): add missing type_vocab_size parameter to TvpConfig (#42928)

  Show description for b62e5b3
  

  majiayu000

  and

  claude

  authored
  b62e5b3

  Copy full SHA for b62e5b3

• [Quantization] Removing misleading int8 quantization in Finegrained FP8 (#42945)

  Show description for 4dc60fd
  MekkCyber

  authored
  4dc60fd

  Copy full SHA for 4dc60fd

• Fp8 dq (#42926)

  Show description for af91c0b
  SunMarc

  authored
  af91c0b

  Copy full SHA for af91c0b

• Add buffers to `_init_weights` for ALL models (#42309)

  Show description for 537c2e3
  Cyrilvallez

  authored
  537c2e3

  Copy full SHA for 537c2e3

• rewrite _process_parameter_type in auto_docstring.py to improve usability (#42431)

  Show description for 728f34c
  

  Yacklin

  and

  Rocketknight1

  authored
  728f34c

  Copy full SHA for 728f34c

• Update `param_element_size` (#42818)

  Show description for dd8057a
  SunMarc

  authored
  dd8057a

  Copy full SHA for dd8057a

• [Tests] Fix CompressedTensors tests (#42935)

  Show description for 0a84654
  

  kylesayrs

  and

  MekkCyber

  authored
  0a84654

  Copy full SHA for 0a84654

• [CB] Allow block sharing in hybrid models (#42877)

  Show description for 04e78e6
  remi-or

  authored
  04e78e6

  Copy full SHA for 04e78e6

• [Quantization] Misc tests fixes (#42940)

  Show description for d7dd443
  MekkCyber

  authored
  d7dd443

  Copy full SHA for d7dd443

• [Quantization] rm _pre_quantization_dtype from quantization tests (#42939)

  Show description for d3d4b62
  MekkCyber

  authored
  d3d4b62

  Copy full SHA for d3d4b62

• 🚨 Generation config defaults are now `None` (#42702)

  Show description for a81e04a
  

  zucchini-nlp

  and

  vasqu

  authored
  a81e04a

  Copy full SHA for a81e04a

• Document new default shard size + dropped unsafe serialization (#42904)

  Show description for bdaddb6
  

  Wauplin

  and

  Cyrilvallez

  authored
  bdaddb6

  Copy full SHA for bdaddb6

Commits on Dec 17, 2025

• Improve BatchFeature (.to() works on lists/nested lists of tensors, auto-skip non array conversion) (#42884)

  Show description for 9f583b1
  yonigozlan

  authored
  9f583b1

  Copy full SHA for 9f583b1

• [docs] optimization cleanup (#42827)

  Show description for 5ef16ed
  stevhliu

  authored
  5ef16ed

  Copy full SHA for 5ef16ed

• [docs] Update shard size (#42749)

  Show description for 12fe95f
  stevhliu

  authored
  12fe95f

  Copy full SHA for 12fe95f

• [docs] Expert parallelism (#42409)

  Show description for 9e3568e
  stevhliu

  authored
  9e3568e

  Copy full SHA for 9e3568e

• [docs] WeightConverter (#42636)

  Show description for 1aab1e9
  stevhliu

  authored
  1aab1e9

  Copy full SHA for 1aab1e9

• docs: Squared ReLU paper fix (#42931)

  Show description for 2f9e21f
  casinca

  authored
  2f9e21f

  Copy full SHA for 2f9e21f

• Remove ipex/ccl in CPU training doc (#42866)

  Show description for f404f15
  

  jiqing-feng

  and

  stevhliu

  authored
  f404f15

  Copy full SHA for f404f15

• Make gradient-checkpoint enabling tolerant of models without get_input_embeddings (#42558)

  Show description for b712a97
  molbap

  authored
  b712a97

  Copy full SHA for b712a97

• Fix dtype quantizer (#42882)

  Show description for b05d2c4
  

  SunMarc

  and

  MekkCyber

  authored
  b05d2c4

  Copy full SHA for b05d2c4

Pagination

Previous

Next