Skip to content

Pull requests: NVIDIA-NeMo/RL

New pull request New
317 Open 1,894 Closed
317 Open 1,894 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

refactor(policy): migrate common config leaves to BaseModel CI:L1 Run doctests, unit tests, and functional tests
#3197 opened Jul 14, 2026 by NolenLiang Contributor Loading…
2
ci: Bump Megatron-Bridge to 482c727 CI:L1 Run doctests, unit tests, and functional tests
#3196 opened Jul 14, 2026 by svcnvidia-nemo-ci Contributor Loading…
4
fix: Fix mtp metrics scale and mtp grad norm logging CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version) r0.7.0
#3194 opened Jul 14, 2026 by yfw Contributor Loading…
4 tasks
2
feat(sglang): megatron backend weight refit for sglang rollouts community-request
#3190 opened Jul 14, 2026 by xiuhu17 Contributor Loading…
1
fix(sglang): avoid ncclCommSplit hang in the refit weight-update group community-request
#3189 opened Jul 14, 2026 by xiuhu17 Contributor Loading…
3
feat(sglang): mxfp8 quantized rollout and refit community-request
#3188 opened Jul 14, 2026 by xiuhu17 Contributor Loading…
1
feat(sglang): engine fault tolerance for rollouts community-request
#3187 opened Jul 14, 2026 by xiuhu17 Contributor Loading…
1
[DO NOT MERGE] NCCL Communication Group Shrink Demo. Documentation Improvements or additions to documentation
#3186 opened Jul 14, 2026 by youngeunkwon0405 Contributor • Draft
4 tasks
1
Add async non-colated PPO + connect to Gym rollout Documentation Improvements or additions to documentation
#3185 opened Jul 13, 2026 by HeyyyyyyG Contributor • Draft
4 tasks
1
ci: add JOB_REAPER_COMMENT support to tools/launch and super-rlvr recipe
#3184 opened Jul 13, 2026 by kajalj22 Contributor Loading…
2 tasks
1
chore: upgrade Python to 3.14.6 CI:L1 Run doctests, unit tests, and functional tests
#3183 opened Jul 13, 2026 by kajalj22 Contributor • Draft
2 tasks
3
fix: handle inflated packed sequence lengths in input-id packing CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version)
#3182 opened Jul 13, 2026 by macandro96 Contributor Loading…
4 tasks
2
feat: Adding fast token CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version)
#3181 opened Jul 13, 2026 by arnavk-nvidia Contributor Loading…
4 tasks
7
[WIP] feat: PPO separate rollout cluster Documentation Improvements or additions to documentation
#3177 opened Jul 13, 2026 by bg51717 Contributor • Draft
1
fix: fix rollout metrics fail in unit test CI:L1 Run doctests, unit tests, and functional tests
#3173 opened Jul 13, 2026 by yuki-97 Contributor Loading…
3
fix: add vLLM config context during collective refit CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version)
#3168 opened Jul 11, 2026 by snowmanwwg Contributor Loading…
4
Vllm diffusion backend
#3167 opened Jul 11, 2026 by linnanwang Loading…
4 tasks
1
feat(megatron): add nccl refit backend for non-colocated generation
#3166 opened Jul 11, 2026 by shanmugamr1992 Contributor Loading…
1 of 2 tasks
1
feat: tq train prefetch
#3162 opened Jul 11, 2026 by zyzhou5 Contributor Loading…
4 tasks
1
feat: log reward component metrics
#3161 opened Jul 10, 2026 by cmunley1 Contributor • Draft
2
fix(vllm): assign node-local VLLM_PORT for cross-node model parallelism CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version)
#3159 opened Jul 10, 2026 by ananthsub Contributor Loading…
3 of 4 tasks
2
perf(ray.sub): tune GCS for large-scale multi-node init bursts CI:docs Run doctest
#3158 opened Jul 10, 2026 by ananthsub Contributor Loading…
7
feat: fault-tolerance checkpoint retention (ft_save_period / ft_keep_latest_k) CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version)
#3157 opened Jul 10, 2026 by ananthsub Contributor Loading…
3 of 4 tasks
10
fix(automodel): avoid fp32 weight load when optimizer holds master community-request waiting-on-maintainers Waiting on maintainers to respond
#3156 opened Jul 10, 2026 by lonexreb Contributor Loading…
1 of 2 tasks
1
2
fix(async): Always honor recompute_kv_cache_after_weight_updates accuracy 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 r0.7.0
#3152 opened Jul 10, 2026 by jepio Loading…
3 of 4 tasks
11
ProTip! Find all pull requests that aren't related to any open issues with -linked:issue.