Commits

Breadcrumbs

History for

onmain

User selector

All users

Datepicker

All time

Commit history

Commits on Jan 14, 2026

• GRPO Trainer: accommodations for reward logging for eval as well as abiulity to not compute eval losses.

  

  M I

  committed
  67582f2View commit details

  Copy full SHA for 67582f2

  View code at this pointBrowse repository at this point

Commits on Oct 23, 2025

• factor out the logps and importance calcualtions from _generate_and_score to _compute_importances

  

  M I

  committed
  be1ea63View commit details

  Copy full SHA for be1ea63

  View code at this pointBrowse repository at this point

Commits on Oct 22, 2025

• Merge remote-tracking branch 'upstream/v0.24-release'

  

  M I

  committed
  1f7209bView commit details

  Copy full SHA for 1f7209b

  View code at this pointBrowse repository at this point

• fix a typo

  

  M I

  committed
  ef32cb1View commit details

  Copy full SHA for ef32cb1

  View code at this pointBrowse repository at this point

Commits on Oct 16, 2025

• Deprecate `BestOfNSampler` (#4291)

  Show description for 19d2f97
  

  qgallouedec

  and

  behroozazarkhalili

  authored
  19d2f97View commit details

  Copy full SHA for 19d2f97

  View code at this pointBrowse repository at this point

• Add accuracy reward (#4270)

  Show description for 8e2d551
  

  pramodith

  and

  qgallouedec

  authored
  8e2d551View commit details

  Copy full SHA for 8e2d551

  View code at this pointBrowse repository at this point

Commits on Oct 15, 2025

• Add support for `token_type_ids` in `DPOTrainer` (#4285)

  Show description for 26b7c25
  

  aweers

  and

  qgallouedec

  authored
  26b7c25View commit details

  Copy full SHA for 26b7c25

  View code at this pointBrowse repository at this point

• Fix CI dev test TypeError: unexpected keyword argument 'load_in_4bit' (#4262)

  albertvillanova

  authored
  7e0adbcView commit details

  Copy full SHA for 7e0adbc

  View code at this pointBrowse repository at this point

• 💰 `RichProgressCallback` enhancement (#4245)

  qgallouedec

  authored
  773afd9View commit details

  Copy full SHA for 773afd9

  View code at this pointBrowse repository at this point

• Fix docstrings with Sphinx 'deprecated' directive (#4279)

  albertvillanova

  authored
  927cf6bView commit details

  Copy full SHA for 927cf6b

  View code at this pointBrowse repository at this point

Commits on Oct 14, 2025

• Fix style with make precommit (#4265)

  albertvillanova

  authored
  cefbacbView commit details

  Copy full SHA for cefbacb

  View code at this pointBrowse repository at this point

• Use FutureWarning instead of DeprecationWarning (#4266)

  albertvillanova

  authored
  fae245aView commit details

  Copy full SHA for fae245a

  View code at this pointBrowse repository at this point

Commits on Oct 13, 2025

• Fix docstring interlinks (#4221)

  albertvillanova

  authored
  2aa9506View commit details

  Copy full SHA for 2aa9506

  View code at this pointBrowse repository at this point

• Raise deprecation warning for Python 3.9 (#4226)

  albertvillanova

  authored
  d6eeb29View commit details

  Copy full SHA for d6eeb29

  View code at this pointBrowse repository at this point

Commits on Oct 10, 2025

• Include `chat_template_kwargs` in `apply_chat_template` (#4233)

  Show description for aab21eb
  

  cmpatino

  and

  qgallouedec

  authored
  aab21ebView commit details

  Copy full SHA for aab21eb

  View code at this pointBrowse repository at this point

• [Online-DPO] fix the completion_len == max_new_tokens crash (#4193)

  Show description for b997a31
  

  3 peopleauthored
  b997a31View commit details

  Copy full SHA for b997a31

  View code at this pointBrowse repository at this point

• Deprecate unused dataset_formatting module (#4242)

  Show description for 039d526
  

  behroozazarkhalili

  and

  qgallouedec

  authored
  039d526View commit details

  Copy full SHA for 039d526

  View code at this pointBrowse repository at this point

• 🧺 [3/N] Refactor `_generate` in GRPO/RLOO: Rely on generator for prompt truncation (#4153)

  qgallouedec

  authored
  0e57b4aView commit details

  Copy full SHA for 0e57b4a

  View code at this pointBrowse repository at this point

• Fix CI slow test ValueError: Backward pass should have cleared tracker of all tensors (#4236)

  Show description for 803ec0d
  

  sywangyi

  and

  kashif

  authored
  803ec0dView commit details

  Copy full SHA for 803ec0d

  View code at this pointBrowse repository at this point

Commits on Oct 9, 2025

• Warnings pointing to RFC (#4224)

  qgallouedec

  authored
  7a0a615View commit details

  Copy full SHA for 7a0a615

  View code at this pointBrowse repository at this point

Commits on Oct 8, 2025

• Remove unused Path import in __init__.py (#4227)

  albertvillanova

  authored
  f784632View commit details

  Copy full SHA for f784632

  View code at this pointBrowse repository at this point

• Fix callable annotations (#4216)

  albertvillanova

  authored
  a944890View commit details

  Copy full SHA for a944890

  View code at this pointBrowse repository at this point

• 🏷️ Account for `token_type_ids` in `DataCollatorForVisionLanguageModeling` (#4190)

  qgallouedec

  authored
  d1d0407View commit details

  Copy full SHA for d1d0407

  View code at this pointBrowse repository at this point

• Fix entropy and accuracy calculation for prompt_tuning techniques. (#4196)

  pramodith

  authored
  f15399dView commit details

  Copy full SHA for f15399d

  View code at this pointBrowse repository at this point

Commits on Oct 7, 2025

• 🧺 [2/N] Refactor `_generate` in GRPO/RLOO: Use `prompt_ids` from generation (#4152)

  qgallouedec

  authored
  cc578b6View commit details

  Copy full SHA for cc578b6

  View code at this pointBrowse repository at this point

• 🎨 Support mixing image+text and text-only examples (#4203)

  Show description for 30cf68a
  

  qgallouedec

  and

  sergiopaniego

  authored
  30cf68aView commit details

  Copy full SHA for 30cf68a

  View code at this pointBrowse repository at this point

• Fix prompt-completion labeling with add_generation_prompt and warning (#4201)

  Show description for 5d87095
  

  behroozazarkhalili

  and

  qgallouedec

  authored
  5d87095View commit details

  Copy full SHA for 5d87095

  View code at this pointBrowse repository at this point

Commits on Oct 6, 2025

• Apply style and revert change in `sft_video_llm` example (#4214)

  qgallouedec

  authored
  65eb45cView commit details

  Copy full SHA for 65eb45c

  View code at this pointBrowse repository at this point

• Removed tokenizer/processor creation from example scripts (#4211)

  sergiopaniego

  authored
  ae6837fView commit details

  Copy full SHA for ae6837f

  View code at this pointBrowse repository at this point

• Remove `Optional` from `processing_class` in `PPOTrainer` (#4212)

  sergiopaniego

  authored
  5291015View commit details

  Copy full SHA for 5291015

  View code at this pointBrowse repository at this point

• Remove tokenizer creation from `sft` example script (#4197)

  sergiopaniego

  authored
  7ad9ce8View commit details

  Copy full SHA for 7ad9ce8

  View code at this pointBrowse repository at this point

Commits on Oct 1, 2025

• add customization to the vllm base class

  

  M I

  committed
  28cdf3aView commit details

  Copy full SHA for 28cdf3a

  View code at this pointBrowse repository at this point

• 🌡️ Have vLLM return processed (temperature scaled) log probs (#4163)

  Show description for e086f07
  

  YonatanGideoni

  and

  qgallouedec

  authored
  e086f07View commit details

  Copy full SHA for e086f07

  View code at this pointBrowse repository at this point

• Hotfix wrong formatting of docstrings with blockquote tips (#4187)

  albertvillanova

  authored
  e5d437eView commit details

  Copy full SHA for e5d437e

  View code at this pointBrowse repository at this point

• Fix handling of f_divergence_type in DPO (#4171)

  Show description for 5a4021f
  

  albertvillanova

  and

  qgallouedec

  authored
  5a4021fView commit details

  Copy full SHA for 5a4021f

  View code at this pointBrowse repository at this point