Skip to content

Pull requests: NVIDIA-NeMo/RL

New pull request New
415 Open 2,143 Closed
415 Open 2,143 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

Port GDPO length adjustment features to GitHub main
#3604 opened Aug 12, 2026 by arendu Contributor • Draft
4 tasks
1
fix: Add torch-memory-saver to the mcore extra deps CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version)
#3602 opened Aug 12, 2026 by tdene Contributor Loading…
4 tasks
2
Update MLPerf Training GRPO dataset community-request
#3601 opened Aug 12, 2026 by filaretov Loading…
1
fix: stop the launcher environment leaking into checkpoint config.yaml community-request
#3600 opened Aug 12, 2026 by mario-at-intercom Loading…
3 of 4 tasks
1
1
fix(async-grpo): eliminate prompt skips across checkpoint save/restore CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version) Documentation Improvements or additions to documentation
#3599 opened Aug 12, 2026 by yfw Contributor • Draft
4 tasks
1
ci: Bump Megatron-Bridge to 600d069 CI:L1 Run doctests, unit tests, and functional tests
#3597 opened Aug 12, 2026 by svcnvidia-nemo-ci Contributor Loading…
3
feat: add DeepSeek V4 Flash support - Automodel path Documentation Improvements or additions to documentation
#3595 opened Aug 12, 2026 by zpqiu Contributor • Draft
4 tasks
1
feat(sc): restart dead generation shards and re-admit them at the next refit
#3592 opened Aug 12, 2026 by terrykong Collaborator • 4/4 Loading…
3 of 4 tasks
2
feat(sc): keep training when a generation shard dies
#3591 opened Aug 12, 2026 by terrykong Collaborator • 3/4 Loading…
3 of 4 tasks
2
feat(sc): track generation fleet health and route NeMo-Gym through NeMo-RL
#3590 opened Aug 12, 2026 by terrykong Collaborator • 2/4 Loading…
3 of 4 tasks
2
feat(sc): contain rollout failures in the SingleController path
#3589 opened Aug 12, 2026 by terrykong Collaborator • 1/4 Loading…
3 of 4 tasks
2
fix: Update comments to include implementation details link CI:docs Run doctest
#3588 opened Aug 12, 2026 by youngeunkwon0405 Contributor Loading…
4 tasks
@youngeunkwon0405
3
perf(external-vllm): remove startup preflights CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version)
#3587 opened Aug 11, 2026 by yfw Contributor Loading…
4 tasks
2
Nano3.5 v2 launch Documentation Improvements or additions to documentation
#3586 opened Aug 11, 2026 by mehraakash Contributor • Draft
4 tasks
1
feat(rollout): add partial recovery benchmark telemetry
#3585 opened Aug 11, 2026 by macandro96 Contributor • Draft
1
feat(opd): add student top-k on-policy distillation CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version) Documentation Improvements or additions to documentation
#3584 opened Aug 11, 2026 by macandro96 Contributor Loading…
4 tasks
2
feat(grpo): optionally skip replay buffer restore CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version)
#3583 opened Aug 11, 2026 by yfw Contributor Loading…
4 tasks
2
feat(sc): add ready-first staleness sampler
#3582 opened Aug 11, 2026 by haitian-nvidia Loading…
3 of 4 tasks
5
perf(vllm): avoid repeated multi-turn tokenization CI:L1 Run doctests, unit tests, and functional tests
#3581 opened Aug 11, 2026 by jthomson04 Contributor Loading…
4 tasks done
4
docs: fix references that point at files which do not exist
#3579 opened Aug 11, 2026 by terrykong Collaborator Loading…
4 tasks done
1
perf(data): embed images in data URLs without re-encoding
#3576 opened Aug 11, 2026 by rohitrango Contributor Loading…
3 of 4 tasks
2
chore: [super-3.5] use custom vllm and flashinfer CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version)
#3575 opened Aug 10, 2026 by guyueh1 Contributor Loading…
4 tasks
5
build(container): support non-root runtime execution CI:L0 Run doctests and unit tests CI Relating to CI
#3574 opened Aug 10, 2026 by ko3n1g Contributor • Draft
3 of 4 tasks
1
feat(profiling): add pluggable policy and vLLM rollout hooks CI:L1 Run doctests, unit tests, and functional tests Documentation Improvements or additions to documentation
#3573 opened Aug 10, 2026 by minitu • Draft
6
fix(megatron): shard Megatron rollouts across DP coordinator leaders community-request
#3571 opened Aug 10, 2026 by YigongQin • Draft
3 of 4 tasks
1
ProTip! Add no:assignee to see everything that’s not assigned.