Skip to content

Pull requests: NousResearch/atropos

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: count other envs' queued data in /status-env when group size is 1
#488 opened Jul 1, 2026 by golldyck Loading…
10 tasks done
fix: remove double retry on chat_completion (9 attempts instead of 3)
#487 opened Jul 1, 2026 by golldyck Loading…
10 tasks done
fix: avoid duplicate request for n=1 when the API ignores n
#486 opened Jul 1, 2026 by golldyck Loading…
10 tasks done
fix: apply sub-reward weights in CombinedReward
#485 opened Jul 1, 2026 by golldyck Loading…
10 tasks done
fix: don't let response-body read mask the original HTTP error
#484 opened Jul 1, 2026 by golldyck Loading…
10 tasks done
fix: resolve correct reward class for modules defining several rewards
#482 opened Jul 1, 2026 by golldyck Loading…
10 tasks done
fix: clamp variance to keep GRPO std real for constant float rewards
#481 opened Jul 1, 2026 by golldyck Loading…
10 tasks done
docs: add WSL troubleshooting for venv setup and pip install memory e…
#470 opened Apr 25, 2026 by PhantomTee Loading…
8 of 12 tasks
fix: Make student/teacher model and tokenizer configurable
#469 opened Apr 24, 2026 by RUFFY-369 Loading…
9 tasks done
fix: Harden process termination with PID verification
#468 opened Apr 24, 2026 by RUFFY-369 Loading…
8 tasks done
fix: Add rollout queue limit and backpressure to prevent OOM
#467 opened Apr 24, 2026 by RUFFY-369 Loading…
8 tasks done
fix: Add bridge config cleanup to prevent stale CUDA IPC leaks
#466 opened Apr 24, 2026 by RUFFY-369 Loading…
9 tasks done
fix: Implement robust advantage normalization for low-variance groups
#465 opened Apr 24, 2026 by RUFFY-369 Loading…
8 tasks done
fix: Implement on-policy teacher distillation with KL loss
#464 opened Apr 24, 2026 by RUFFY-369 Loading…
9 tasks done
fix: Enhance RoPE theta detection and fix meta tensor traversal
#463 opened Apr 24, 2026 by RUFFY-369 Loading…
9 tasks done
fix: Implement strict dtype validation for shared vLLM tensors
#462 opened Apr 24, 2026 by RUFFY-369 Loading…
9 tasks done
OpenReward Integration
#451 opened Apr 22, 2026 by RUFFY-369 Loading…
8 of 9 tasks
ProTip! Updated in the last three days: updated:>2026-06-29.