Skip to content

Pull requests: volcengine/verl

New pull request New
189 Open 1,893 Closed
189 Open 1,893 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

Revert "[trainer] fix: address serialization issues when using async reward function and ray ppo trainer"
#3819 opened Oct 18, 2025 by vermouth1992 Loading…
2
[trainer] fix: Add data.seed to config
#3815 opened Oct 18, 2025 by HollowMan6 Loading…
7 tasks done
1
13
[data] feat: filter out malformed data together with long prompts
#3814 opened Oct 18, 2025 by HollowMan6 Loading…
7 tasks done
6
[data, trainer] feat: add support for limiting samples from dataset
#3812 opened Oct 18, 2025 by HollowMan6 Loading…
7 tasks done
33
Fix: Remove torch.quantile-based percentile metrics to resolve tensor size limit error
#3810 opened Oct 18, 2025 by szrlee Loading…
1
2
fix: gradient vanish in FusedLinearForPPO
#3765 opened Oct 14, 2025 by sanxing-chen Loading…
1 of 7 tasks
2
[WIP] feat: support humanline for GRPO/PPO trainer
#3751 opened Oct 13, 2025 by sijial430 • Draft
build(deps): bump sglang[all] from 0.5.2 to 0.5.3.post1 dependencies Pull requests that update a dependency file python Pull requests that update python code
#3749 opened Oct 13, 2025 by dependabot bot Loading…
[recipe, hardware] feat: Add GRPO with full weight updates for 1.5B models on a single GPU
#3747 opened Oct 13, 2025 by KabakaWilliam Loading…
6 tasks done
6
[trainer] feat: vlm support for sft engine
#3729 opened Oct 11, 2025 by techkang Loading…
8
[recipe] [WIP] feat: TPPO [passed continuous running] [accuracy unverified]
#3728 opened Oct 11, 2025 by HanlinDu • Draft
7 tasks
[model]{bug-fix}: Fix wrong log prob calculations using fused kernels and USP
#3724 opened Oct 10, 2025 by joshyan1 Loading…
4 of 7 tasks
3
[WIP] [single_controller] feat: PyTorch Monarch integration
#3713 opened Oct 9, 2025 by keyan • Draft
9 tasks
7
Feat/log fit sub reward
#3710 opened Oct 9, 2025 by Arinlzy Loading…
7 tasks
2
[trainer] fix: data balance dp tokens logic with sequence parallelism
#3691 opened Oct 7, 2025 by puneeshkhanna Loading…
3 of 7 tasks
3
Eval protocol integration
#3690 opened Oct 6, 2025 by benjibc • Draft
[data] fix: Remove duplicated bos tokens in RLHFDataset
#3682 opened Oct 5, 2025 by kAIto47802 Loading…
7 tasks done
3
[trainer, worker] feat: more flexible and easy-to-use reward model
#3679 opened Oct 4, 2025 by yyDing1 Loading…
35
fix Ray runtime env no working dir issue
#3673 opened Oct 4, 2025 by kunling-anyscale Loading…
7 tasks
3
[BREAKING][misc] feat: Abstract optimizer
#3656 opened Sep 30, 2025 by EduardDurech Loading…
8
retain origin dtype in compute log prob for megatron backend
#3651 opened Sep 30, 2025 by jiaqiw09 • Draft
4 of 7 tasks
1
[data] feat: TransferQueue - An asynchronous streaming data management system
#3649 opened Sep 30, 2025 by 0oshowero0 Loading…
7 tasks done
7
[algo] feat: Add SVD-LoRA GRPO
#3637 opened Sep 27, 2025 by alekseymalakhov11 Loading…
5 of 7 tasks
8
[recipe] feat: Qwen3-235B-A22B on Ascend NPU
#3628 opened Sep 25, 2025 by johnjunjun7 Loading…
7 tasks
7
[recipe] An agent-lightning like RL training pipeline
#3610 opened Sep 25, 2025 by linxxx3 Loading…
7 tasks done
9
ProTip! Type g i on any issue or pull request to go back to the issue listing page.