Skip to content

Pull requests: huggingface/trl

New pull request New
174 Open 3,641 Closed
174 Open 3,641 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

Fix DPO/KTO evaluate crash with a dict eval_dataset and precompute_ref_log_probs
#6370 opened Jul 13, 2026 by albertvillanova Member Loading…
4
Stop overcounting dead groups and increase your accuracy
#6368 opened Jul 13, 2026 by SushantDaga Loading…
4 of 8 tasks
1
3
Avoid in-place mutation of input prompts in DPPO environments loop
#6366 opened Jul 12, 2026 by albertvillanova Member Loading…
4
Avoid in-place mutation of input prompts in GRPO environments loop
#6364 opened Jul 12, 2026 by albertvillanova Member Loading…
5
Fix corrupted reasoning blocks in Qwen3-family training chat templates
#6363 opened Jul 12, 2026 by qgallouedec Member Loading…
1
1
Add security policy to the repo
#6362 opened Jul 11, 2026 by albertvillanova Member Loading…
12
Bump the actions group with 9 updates dependencies Pull requests that update a dependency file github_actions Pull requests that update GitHub Actions code
#6360 opened Jul 11, 2026 by dependabot Bot Loading…
@albertvillanova
1
Truncate GOLD on-policy prompts before generation, keeping the prompt end
#6359 opened Jul 11, 2026 by roycho96 Contributor Loading…
3 tasks done
@kashif @cmpatino
2
Add trl.environments submodule with SandboxEnvironment
#6358 opened Jul 11, 2026 by qgallouedec Member • Draft
Fix EOS being masked when pad_token_id equals eos_token_id in GKD and GOLD
#6357 opened Jul 11, 2026 by roycho96 Contributor Loading…
3 tasks done
@kashif @cmpatino
Force use_reentrant=True for PEFT + ZeRO-3 + gradient checkpointing in all trainers
#6356 opened Jul 10, 2026 by qgallouedec Member Loading…
1
3
Add support for vLLM 0.24.0
#6355 opened Jul 10, 2026 by qgallouedec Member Loading…
4
Support iterable datasets in GRPO and RLOO
#6351 opened Jul 10, 2026 by albertvillanova Member Loading…
1
9
Environment-owned dataset: make train_dataset optional
#6349 opened Jul 10, 2026 by qgallouedec Member Loading…
1
14
Add LoRA support for AsyncGRPO
#6344 opened Jul 9, 2026 by jonahsamost Loading…
1 of 8 tasks
@AmineDiro
11
Preserve model forward signature in the chunked CE lm_head patch
#6342 opened Jul 9, 2026 by sergiopaniego Member Loading…
3 of 6 tasks
3
Fix SDPO scale_rewards CLI parsing and warn on ignored feedback
#6341 opened Jul 9, 2026 by sergiopaniego Member Loading…
3 of 6 tasks
1
Test KTO training on iterable datasets with the KL loss
#6338 opened Jul 9, 2026 by albertvillanova Member Loading…
1
Test all supported eval_dataset input types in core trainer init tests
#6336 opened Jul 9, 2026 by albertvillanova Member Loading…
1
Fix incorrect arXiv link for the RLOO paper in the KL-penalty comment
#6332 opened Jul 9, 2026 by DaoyuanLi2816 Contributor Loading…
2 tasks done
Fix cast_lm_head_to_fp32 under FSDP2 / DeepSpeed ZeRO-3
#6330 opened Jul 9, 2026 by qgallouedec Member Loading…
1
7
[GOLD] Tool calling support for GOLDTrainer
#6328 opened Jul 8, 2026 by Strongich Contributor Loading…
3 of 6 tasks
8
pip install -e ".[dev]" fails on macOS: liger-kernel requires triton
#6327 opened Jul 8, 2026 by clementtisseau Loading…
2 of 8 tasks
1
Drop examples with no completion left after truncation in DPO and KTO
#6321 opened Jul 7, 2026 by qgallouedec Member Loading…
4
Drop fully-masked examples after truncation in SFT
#6320 opened Jul 7, 2026 by qgallouedec Member Loading…
1
1
ProTip! Filter pull requests by the default branch with base:main.