Skip to content

Pull requests: nex-agi/NexRL

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
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort

Pull requests list

fix: auto-install Weaver SDK for service training
#57 opened May 13, 2026 by CriusT Collaborator Loading…
[fix]Fix train_service legacy detection when role is set
#56 opened May 3, 2026 by CriusT Collaborator Loading…
Propagate sampling filters to Weaver inference
#55 opened Apr 27, 2026 by CriusT Collaborator Loading…
[fix] Ensure weaver client close() on process exit
#54 opened Apr 17, 2026 by CriusT Collaborator Loading…
4 tasks
Skip initial weight sync when training from scratch
#53 opened Apr 16, 2026 by CriusT Collaborator Loading…
4 tasks
Support inference old logprobs in RL training
#52 opened Mar 23, 2026 by CriusT Collaborator Loading…
feat: wire sampling_mask and old_logprob flags end-to-end
#51 opened Mar 23, 2026 by CriusT Collaborator Loading…
1 of 3 tasks
feat: dynamic student-to-teacher weight transfer in GRPO-KL (#140)
#50 opened Mar 20, 2026 by CriusT Collaborator Loading…
4 tasks
[feat] Add MathBoxedLLMJudgeFn and configurable LLM judge API key
#49 opened Mar 18, 2026 by CriusT Collaborator Loading…
3 tasks done
Add DeepSeek V3 thinking chat process_func for StreamingDataset
#48 opened Mar 18, 2026 by CriusT Collaborator Loading…
4 tasks
ProTip! Add no:assignee to see everything that’s not assigned.