Skip to content

Pull requests: huggingface/trl

New pull request New
181 Open 4,271 Closed
181 Open 4,271 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

[Exploratory] Drop accelerate launch: the CLI, docs and examples launch with torchrun
#7180 opened Sep 10, 2026 by qgallouedec Member • Draft
Add support for vLLM 0.29.0
#7178 opened Sep 10, 2026 by qgallouedec Member Loading…
2
Remove outdated "under construction" banners from docs
#7177 opened Sep 10, 2026 by qgallouedec Member Loading…
2
AsyncGRPO: cancel stale in-flight rollouts and survive generation fai…
#7176 opened Sep 10, 2026 by AmineDiro Member Loading…
4
[AsyncGRPO] Run sync tools on a thread pool and support async tools
#7175 opened Sep 10, 2026 by AmineDiro Member Loading…
5
Key the doc build concurrency group on the pull request number
#7174 opened Sep 10, 2026 by albertvillanova Member Loading…
1
Cancel superseded runs on main
#7172 opened Sep 10, 2026 by albertvillanova Member • 2/2 Loading…
1
Skip the tests on pushes that change no tested file
#7171 opened Sep 10, 2026 by albertvillanova Member Loading…
1
Cancel superseded PR runs in the test workflows
#7170 opened Sep 10, 2026 by albertvillanova Member • 1/2 Loading…
5
Section the auto-generated release notes by label
#7166 opened Sep 10, 2026 by albertvillanova Member Loading…
2 tasks
1
Publish to PyPI on tag push and guard VERSION against the tag
#7165 opened Sep 10, 2026 by albertvillanova Member Loading…
1
fix: validate fractions in get_dataset to prevent ZeroDivisionError
#7164 opened Sep 10, 2026 by CJstate Loading…
5 of 6 tasks
1
Reject use_liger_kernel in GMPOTrainer
#7163 opened Sep 10, 2026 by YeonwooSung Contributor Loading…
5 of 8 tasks
1
Restore generation unwrap state if generate() raises
#7162 opened Sep 10, 2026 by YeonwooSung Contributor Loading…
4 of 8 tasks
1
Fail fast when GRPO combines Liger with the MoE auxiliary loss
#7161 opened Sep 10, 2026 by YeonwooSung Contributor Loading…
4 of 8 tasks
1
feat(grpo): add compute_metrics support to GRPOTrainer
#7144 opened Sep 9, 2026 by behroozazarkhalili Collaborator Loading…
4 of 5 tasks
1
3
feat: add activation_offloading to GRPOTrainer and RLOOTrainer
#7143 opened Sep 9, 2026 by behroozazarkhalili Collaborator Loading…
4 of 5 tasks
1
1
fix: validate dataset mixture fractions before loading
#7142 opened Sep 9, 2026 by behroozazarkhalili Collaborator Loading…
4 of 5 tasks
1
1
fix: score think_format_reward when <think> is prefilled in the prompt
#7141 opened Sep 9, 2026 by behroozazarkhalili Collaborator Loading…
4 of 5 tasks
1
1
Remove the experimental BCO trainer
#7136 opened Sep 9, 2026 by qgallouedec Member Loading…
4
Remove the experimental PRM trainer
#7133 opened Sep 9, 2026 by qgallouedec Member Loading…
2
[AsyncGRPO] Gather FSDP2 weights before NCCL send to vLLM
#7131 opened Sep 9, 2026 by sfc-gh-truwase Loading…
4 of 8 tasks
1
4
fix: validate ngram_size in get_repetition_penalty_reward
#7129 opened Sep 9, 2026 by CJstate Loading…
5 of 6 tasks
1
perf(utils): vectorise selective_log_softmax for all dtypes
#7125 opened Sep 9, 2026 by piyushumate Loading…
4 of 6 tasks
6
New example: [REDACTED]
#7121 opened Sep 9, 2026 by qgallouedec Member Loading…
13
ProTip! Mix and match filters to narrow down what you’re looking for.