Commits

Branch selector

 main

User selector

All users

All time

Commit History

Commits on Feb 24, 2026

• Fix `SFTTrainer` support for single-image data (#5132)

  qgallouedec

  authored
  cfbdd3b

  Copy full SHA for cfbdd3b

Commits on Feb 23, 2026

• Add support for Python 3.14 (#4225)

  Show description for fa313fd
  

  albertvillanova

  and

  qgallouedec

  authored
  fa313fd

  Copy full SHA for fa313fd

• Fix type of TrainingArguments.logging_steps in docs (#5149)

  albertvillanova

  authored
  bc4edf6

  Copy full SHA for bc4edf6

• Use BaseConfig in all experimental configs (#5148)

  albertvillanova

  authored
  5269393

  Copy full SHA for 5269393

• Fix PPOTrainer.save_model (#5151)

  albertvillanova

  authored
  ef08730

  Copy full SHA for ef08730

• Fix wording in DPO and SFT trainer documentation for clarity (#5140)

  qgallouedec

  authored
  ae97f06

  Copy full SHA for ae97f06

• Move common fields from stable trainer configs to BaseConfig (#5136)

  albertvillanova

  authored
  f150780

  Copy full SHA for f150780

Commits on Feb 22, 2026

• refactor(gkd_trainer): small optim (#5143)

  casinca

  authored
  93f2e48

  Copy full SHA for 93f2e48

Commits on Feb 20, 2026

• Add `environment_factory` to `GRPOTrainer` (#5093)

  Show description for 8067ea7
  

  qgallouedec

  and

  sergiopaniego

  authored
  8067ea7

  Copy full SHA for 8067ea7

• Fix `trl <command> --help` TypeError caused by unescaped `%` in `TrainingArguments` help strings (#5135)

  albertvillanova

  authored
  7a4156a

  Copy full SHA for 7a4156a

• Fix NameError: name 'importlib' is not defined (#5134)

  albertvillanova

  authored
  c3ead5b

  Copy full SHA for c3ead5b

• Fix import latency [2/N]: Implement native _is_package_available (#5129)

  albertvillanova

  authored
  b7fa6bf

  Copy full SHA for b7fa6bf

• Fix import latency [1/N]: Extract _LazyModule to dedicated module (#5128)

  albertvillanova

  authored
  bb14764

  Copy full SHA for bb14764

• Refactor DPO (#3906)

  Show description for e3b7897
  

  3 peopleauthored
  e3b7897

  Copy full SHA for e3b7897

Commits on Feb 19, 2026

• Remove revision references in dataset loading for toolcall tests (#5133)

  qgallouedec

  authored
  a68fb89

  Copy full SHA for a68fb89

• Refactor TRL CLI into modular command architecture (#5124)

  albertvillanova

  authored
  699b842

  Copy full SHA for 699b842

• Implement Agent Skills [4/N]: Create skills CLI (#5103)

  albertvillanova

  authored
  b46614e

  Copy full SHA for b46614e

Commits on Feb 18, 2026

• Update tool handling to support JSON string schemas in trainers (#5118)

  qgallouedec

  authored
  f818188

  Copy full SHA for f818188

• Add Tiny Aya tool calling examples (script/notebook) (#5123)

  Show description for 9fc9a7d
  

  sergiopaniego

  and

  qgallouedec

  authored
  9fc9a7d

  Copy full SHA for 9fc9a7d

• Add GLM-4.5 model to tests (#5114)

  qgallouedec

  authored
  2743113

  Copy full SHA for 2743113

• Add check for `None` in `get_trackio_space_url()` to prevent errors (#5115)

  qgallouedec

  authored
  0e531bd

  Copy full SHA for 0e531bd

• Fix Qwen3 schema (#5111)

  qgallouedec

  authored
  8b082bb

  Copy full SHA for 8b082bb

• Add test for Cohere2 models (#5116)

  qgallouedec

  authored
  269217f

  Copy full SHA for 269217f

• Add more tests for `get_training_chat_template` (#5108)

  qgallouedec

  authored
  57df014

  Copy full SHA for 57df014

Commits on Feb 17, 2026

• Update version check for transformers to 5.2.0 in online_dpo_trainer.py (#5110)

  qgallouedec

  authored
  70efa96

  Copy full SHA for 70efa96

• Add validation for conversational prompts in multimodal training (#5067)

  qgallouedec

  authored
  269ed99

  Copy full SHA for 269ed99

• Implement Agent Skills [3/N]: Create skills installer (#5100)

  albertvillanova

  authored
  997536a

  Copy full SHA for 997536a

Commits on Feb 16, 2026

• Remove outdated liger-kernel compatibility checks and warnings in tests and SFTTrainer (#5105)

  qgallouedec

  authored
  8b9b972

  Copy full SHA for 8b9b972

• Fix SFT loss type rewards being overwritten in dpo_loss() (#5079)

  Mr-Neutr0n

  authored
  8c232f6

  Copy full SHA for 8c232f6

• Add Trackio integration for model card visualization (#5101)

  qgallouedec

  authored
  99b26fb

  Copy full SHA for 99b26fb

• Fix style (#5106)

  albertvillanova

  authored
  c94c032

  Copy full SHA for c94c032

• Implement Agent Skills [2/N]: Create skills module (#5097)

  albertvillanova

  authored
  3d1c785

  Copy full SHA for 3d1c785

• feature: top_k selective_log_softmax (#5104)

  Show description for 1702fc0
  

  LeonEricsson

  and

  qgallouedec

  authored
  1702fc0

  Copy full SHA for 1702fc0

• Fix DPO and RLOO incompatibility with FSDP2 (#4838)

  Show description for 29ace1a
  

  flutist

  and

  qgallouedec

  authored
  29ace1a

  Copy full SHA for 29ace1a

• Pass vllm_is_ratio to LigerFusedLinearGRPOLoss in compute_liger_loss (#5031)

  Show description for 3d2e898
  

  yukiu00

  and

  kashif

  authored
  3d2e898

  Copy full SHA for 3d2e898

Pagination

Previous

Next