Skip to content

Pull requests: huggingface/trl

New pull request New
96 Open 2,518 Closed
96 Open 2,518 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

[Experimental] Add SDFT trainer, config, docs, and tests
#4941 opened Jan 31, 2026 by Shekswess Loading…
4 of 5 tasks
1
1
Update RewardFunc type to use RewardCallable protocol
#4938 opened Jan 31, 2026 by amit9oct Loading…
2 of 5 tasks
3
documentation for modifying chat templates for assistant-only loss
#4937 opened Jan 30, 2026 by jiosephlee Loading…
Add Wordle example with Qwen3 thinking activated
#4936 opened Jan 30, 2026 by sergiopaniego • Draft
5 tasks
Add SDPO (Self-Distillation Policy Optimization) trainer
#4935 opened Jan 30, 2026 by MengAiDev Loading…
1
fix: prevent O(2^n) regex backtracking in qwen3_schema
#4934 opened Jan 29, 2026 by wingding12 • Draft
1 task done
1
fix: sanitize malformed tool calls to prevent TypeError in chat templates
#4933 opened Jan 29, 2026 by wingding12 • Draft
3 tasks done
1
Automatically add generation tags to chat template for assistant_only_loss=True training (TRL Issue #4879)
#4900 opened Jan 26, 2026 by Neelectric • Draft
3 of 5 tasks
2
Update wordle.py example with masking of env tokens
#4895 opened Jan 26, 2026 by sergiopaniego Loading…
5 tasks
14
Expose generation index to tool callables in GRPOTrainer
#4894 opened Jan 25, 2026 by lukehinds Loading…
4 tasks done
1
1
Upgrade GitHub Actions to latest versions
#4893 opened Jan 24, 2026 by salmanmkc Loading…
2
[GRPO] feat: Geometric Sequence Masking
#4891 opened Jan 24, 2026 by LeonEricsson Loading…
5 tasks
13
Fix grpo tool calling
#4890 opened Jan 23, 2026 by akshayballal95 Loading…
2 tasks done
1
@qgallouedec
fix(vLLM): Add tool calling support to VLLMClient.chat()
#4889 opened Jan 23, 2026 by kansalaman Loading…
1 of 2 tasks
1
Add History-Aware Adaptive Difficulty Weighting (HA-DW) to GRPO
#4872 opened Jan 20, 2026 by anonx3247 Loading…
2
NeMo-Gym Integration
#4848 opened Jan 17, 2026 by cmunley1 Loading…
28
make dpo compatible with fsdp2
#4838 opened Jan 16, 2026 by flutist Loading…
4 of 5 tasks
feat: Support log_completion for swanlab backend
#4826 opened Jan 14, 2026 by ZiyiTsang Loading…
2 of 5 tasks
1
Add support for training with multiple OpenEnv environments
#4824 opened Jan 13, 2026 by lewtun • Draft
5 tasks
9
Add Entropy Adaptive Fine Tuning to SFT Trainer
#4802 opened Jan 10, 2026 by electroglyph Loading…
1
2
forward_masked_logits in SFTTrainer
#4794 opened Jan 8, 2026 by qgallouedec • Draft
5 tasks
1
Refactor KTO [3/N]: Extract dataset processing to _prepare_dataset method
#4788 opened Jan 8, 2026 by albertvillanova Loading…
3
Refactor KTO [2/N]: Improve config validation in KTOConfig
#4787 opened Jan 8, 2026 by albertvillanova Loading…
4
make dpo compatible with qwen3vl
#4773 opened Jan 4, 2026 by flutist Loading…
17
feat(sft): add generation-based evaluation support to SFTTrainer
#4768 opened Jan 2, 2026 by CodersAcademy006 Loading…
1
5
ProTip! Updated in the last three days: updated:>2026-01-29.