Skip to content

Pull requests: verl-project/verl

New pull request New
314 Open 2,605 Closed
314 Open 2,605 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

[training_utils] fix: fixed various bugs with tracking
#5135 opened Jan 30, 2026 by guillemgt Loading…
3 of 7 tasks
[reward] feat: mv split reward loop manager and agent loop manager
#5134 opened Jan 30, 2026 by yyDing1 Loading…
8 tasks
[megatron] integrate MCore blackwell linear_cross_entropy fusion kernel
#5130 opened Jan 30, 2026 by HaochenYuan Loading…
8 tasks
[data] fix: correct image handling in RLHFDataset
#5126 opened Jan 30, 2026 by Cosmicstring Loading…
4 of 8 tasks
[perf] fix: infer server profiler args fix
#5121 opened Jan 30, 2026 by mengchengTang Loading…
8 tasks
[model] feat: add registration and config converter for Qwen 2.5-Omni
#5120 opened Jan 30, 2026 by martinzhang03 Loading…
2
[model, algo] feat: implement SAC algorithm and support Pi0.5 model
#5118 opened Jan 30, 2026 by Miical Loading…
8 tasks
27
[rollout] fix: remove dtype cast
#5117 opened Jan 30, 2026 by vermouth1992 Loading…
8 tasks
[fsdp] feat: Merge lora in fsdp training to speed up rollout
#5115 opened Jan 30, 2026 by amzfang • Draft
3 of 8 tasks
1
[sglang] Fixing issues in examples/sglang_multiturn/search_r1_like/run_qwen2.5-3b_instruct_search_multiturn.sh
#5114 opened Jan 29, 2026 by NLPJCL Loading…
1
[vllm] fix: ignore MoE router layers for FP8 quantization
#5107 opened Jan 29, 2026 by zpqiu • Draft
8 tasks
[WIP][perf] doc: add perf_tuning_on_ascend
#5104 opened Jan 29, 2026 by tardis-key • Draft
8 tasks
1
[trtllm, rollout] test: add unittest
#5102 opened Jan 29, 2026 by hchings • Draft
8 tasks
[rollout] fix: fix "input_ids in tensor_dict1 and tensor_dict2 are not the same object" when running the DeepEyes in protocol
#5101 opened Jan 29, 2026 by ChibiQuest Loading…
8 tasks
[training_utils] fix: Resolved bugs and conflicts in the fully async caused by multiple PRs
#5100 opened Jan 29, 2026 by ZLiao097 Loading…
4 of 8 tasks
6
[rollout] fix: fix cpu allocation error in tensorrt_llm rollout manager, with more robust option
#5097 opened Jan 29, 2026 by SchumiDing • Draft
7 of 8 tasks
6
[fsdp] test: add backward determinism test
#5096 opened Jan 28, 2026 by Luosuu Loading…
4 of 8 tasks
1
[vllm] fix: build_app() missing 1 required positional argument: 'supported_tasks'
#5093 opened Jan 28, 2026 by HollowMan6 Loading…
8 tasks done
1
[megatron] fix: megatron args for fully async recipe
#5092 opened Jan 28, 2026 by yyDing1 Loading…
8 tasks
[rollout,vllm] feat: add wake_up_replica method
#5086 opened Jan 28, 2026 by RobotGF Loading…
8 tasks
[rollout,vllm] feat: add exceptioin handle for vllm in async mode
#5082 opened Jan 28, 2026 by RobotGF Loading…
8 tasks
4
[fsdp]Optimizing old logp onload/offload with non-blocking copy
#5081 opened Jan 28, 2026 by Shangwei-Li • Draft
4 of 8 tasks
[trainer, rollout] fix: release rollout GPU memory after pre-train validation
#5075 opened Jan 28, 2026 by none0663 Loading…
8 tasks
6
[ray, single_controller, trainer, utils] feat: Add distributed backend selection support (Ray/openYuanrong)
#5067 opened Jan 27, 2026 by mianmianboom Loading…
3 of 8 tasks
2
[BREAKING][megatron] feat: support dynamic CP
#5057 opened Jan 27, 2026 by ISEEKYAN • Draft
ProTip! no:milestone will show everything without a milestone.