-
Notifications
You must be signed in to change notification settings - Fork 400
Pull requests: NVIDIA-NeMo/RL
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
feat(grpo): checkpoint async replay buffer
CI:Lfast
Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version)
#2632
opened May 29, 2026 by
macandro96
Contributor
Loading…
4 tasks
ci: Bump Megatron-Bridge to 44bb82f
CI:L1
Run doctests, unit tests, and functional tests
#2629
opened May 29, 2026 by
svcnvidia-nemo-ci
Loading…
ci: Fix Lfast tests
CI:Lfast
Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version)
CI
Relating to CI
#2628
opened May 29, 2026 by
chtruong814
Contributor
Loading…
4 tasks
docs: add two-stage SWE RL guide and recipes for Qwen3-30B-A3B-Thinking
Documentation
Improvements or additions to documentation
#2624
opened May 29, 2026 by
binhu-nv
Loading…
Add shard-before-load option for DTensor model loading
#2621
opened May 29, 2026 by
Mingyu-Yang-1
Loading…
test: add TQ nightly coverage set (simple + mooncake_cpu backends)
CI:L0
Run doctests and unit tests
#2616
opened May 28, 2026 by
ZhiyuLi-Nvidia
Contributor
Loading…
4 tasks
Fix "No backend type associated with device type cpu" error with dynamic sampling
#2614
opened May 28, 2026 by
ashors1
Contributor
Loading…
4 tasks
feat: Numa aware binding
CI:Lfast
Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version)
#2613
opened May 28, 2026 by
youngeunkwon0405
Contributor
Loading…
4 tasks
feat: Topology aware placement
CI:Lfast
Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version)
#2612
opened May 28, 2026 by
youngeunkwon0405
Contributor
Loading…
4 tasks
fix: update ray executor import for vLLM 0.20
CI:Lfast
Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version)
#2609
opened May 28, 2026 by
achartier
Contributor
Loading…
4 tasks done
feat: add checkpoint-engine refit interface and integrate NIXL
#2608
opened May 28, 2026 by
HollowMan6
Member
•
Draft
3 of 4 tasks
ci(session-memory): NVSkills signing — license, evals, secrets baseline
CI:docs
Run doctest
#2605
opened May 28, 2026 by
terrykong
Collaborator
Loading…
ci(launch-nemo-rl): NVSkills signing — license, evals, secrets baseline
CI:docs
Run doctest
#2604
opened May 28, 2026 by
terrykong
Collaborator
Loading…
ci(docs): NVSkills signing — license, evals, secrets baseline
CI:docs
Run doctest
#2603
opened May 28, 2026 by
terrykong
Collaborator
Loading…
ci(brev-etiquette): NVSkills signing — license, evals, secrets baseline
CI:docs
Run doctest
#2602
opened May 28, 2026 by
terrykong
Collaborator
Loading…
ci(auto-research): NVSkills signing — license, evals, secrets baseline
CI:docs
Run doctest
#2601
opened May 28, 2026 by
terrykong
Collaborator
Loading…
feat: add Claude skill for building new native RL environments
CI:docs
Run doctest
#2598
opened May 28, 2026 by
terrykong
Collaborator
Loading…
3 tasks
refactor: handle GDPO multi-reward by dict instead of positional list
CI:L1
Run doctests, unit tests, and functional tests
Documentation
Improvements or additions to documentation
#2597
opened May 28, 2026 by
NolenLiang
Contributor
Loading…
4 tasks
ci: add evals scaffolding to skills for NVSkills CI signing
CI:docs
Run doctest
#2595
opened May 28, 2026 by
terrykong
Collaborator
Loading…
1 task
test(data_plane): pin async-RL filter flow + example BaseRolloutFilter
CI:L1
Run doctests, unit tests, and functional tests
#2593
opened May 28, 2026 by
ZhiyuLi-Nvidia
Contributor
Loading…
4 tasks done
feat(modelopt): support real NVFP4 QAT rollout
#2592
opened May 27, 2026 by
HollowMan6
Member
Loading…
3 of 4 tasks
fix: disable flashinfer MOE FP16 in container
#2589
opened May 27, 2026 by
kajalj22
Contributor
Loading…
2 tasks
fix(policy): re-onload model on cuda before DTensor v2 weight refit
CI:L2
Run doctests, unit tests, functional tests, and convergence tests
#2587
opened May 27, 2026 by
qiaochuz-nv
Contributor
Loading…
5 tasks done
Previous Next
ProTip!
no:milestone will show everything without a milestone.