Skip to content

Pull requests: NVIDIA-NeMo/RL

New pull request New
566 Open 2,457 Closed
566 Open 2,457 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

fix: Fix vlm_grpo-nemotron-omni-30ba3b nightly
#4134 opened Sep 15, 2026 by tdene Contributor Loading…
4 tasks
1
Zongfeij/dev community-request waiting-on-customer Waiting on the original author to respond
#4131 opened Sep 15, 2026 by zongfeijing Loading…
4 tasks
2
Bxyu/dev 20260908
#4130 opened Sep 14, 2026 by bxyu-nvidia Contributor • Draft
4 tasks
1
feat(rollout): integrate MInf ledger capture with checkpointing CI:L0 Run doctests and unit tests Documentation Improvements or additions to documentation
#4129 opened Sep 14, 2026 by lauradang Loading…
38
Adithyare/super v3.5 joint mopd v4
#4128 opened Sep 14, 2026 by arendu Contributor • Draft
4 tasks
1
feat(megatron): support selective MXFP8 inference parameters CI:L1 Run doctests, unit tests, and functional tests Documentation Improvements or additions to documentation
#4126 opened Sep 14, 2026 by wdykas Contributor • Draft
3
feat(draft) - Add Dynamic CP Support Documentation Improvements or additions to documentation
#4125 opened Sep 14, 2026 by humairafirdowse18 Contributor Loading…
1
feat(cc): segment-aware SingleController training
#4124 opened Sep 14, 2026 by aroshanghias-nvd Contributor • Draft
2 of 6 tasks
1
feat(trtllm): Rubin build support and rollout fixes
#4123 opened Sep 14, 2026 by hchings Contributor • Draft
4 tasks
@hchings
1
feat(trtllm): Qwen3.5 routed-expert block-FP8/MXFP8 refit Documentation Improvements or additions to documentation
#4122 opened Sep 14, 2026 by hchings Contributor • Draft
4 tasks
@hchings
10
fix(grpo): mask excluded values before logprob exponentiation community-request waiting-on-maintainers Waiting on maintainers to respond
#4120 opened Sep 13, 2026 by bzantium Loading…
3 of 4 tasks
1
feat(ppo): add DeepSeek R1 7B single-controller recipe
#4119 opened Sep 13, 2026 by snowmanwwg Contributor Loading…
4
perf: enable HybridEP in remaining GB200 MoE performance recipes CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version)
#4118 opened Sep 12, 2026 by seonjinn Contributor Loading…
4 of 7 tasks
2
feat(checkpoint): Support turn-level checkpoint Documentation Improvements or additions to documentation
#4117 opened Sep 12, 2026 by macandro96 Contributor • Draft
4 tasks
4
fix(megatron): bind gradient callbacks to runtime model config CI:L0 Run doctests and unit tests
#4116 opened Sep 12, 2026 by DanialTaheri Contributor Loading…
3 of 4 tasks
2
feat(automodel): expose FSDP output precision
#4115 opened Sep 12, 2026 by pstjohn Contributor Loading…
2 of 4 tasks
15
fix(automodel): honor training logprob chunks for local vocabulary
#4114 opened Sep 12, 2026 by pstjohn Contributor Loading…
3 of 4 tasks
25
fix(vlm): forward image payloads in async rollouts CI:L2 Run doctests, unit tests, functional tests, and convergence tests
#4113 opened Sep 12, 2026 by jinglinglingling Contributor Loading…
11
fix: build refit metadata before communicator initialization CI:L0 Run doctests and unit tests
#4111 opened Sep 12, 2026 by seonjinn Contributor • Draft
3
fix(vllm): allow BF16 rollout with inherited quantization exclusions CI:L0 Run doctests and unit tests
#4110 opened Sep 12, 2026 by seonjinn Contributor • Draft
4
feat(sft): add Nemotron task encoders and cookers
#4108 opened Sep 12, 2026 by rohitrango Contributor • 4/4 Loading…
2 of 4 tasks
14
perf(data-plane): avoid local batch copies
#4107 opened Sep 12, 2026 by rohitrango Contributor • 3/4 Loading…
3 of 4 tasks
2
feat(data): add PackedTensor preprocessing modes
#4106 opened Sep 12, 2026 by rohitrango Contributor • 2/4 Loading…
6 tasks done
2
feat: add Energon-owned SFT sequence packing Documentation Improvements or additions to documentation
#4105 opened Sep 12, 2026 by rohitrango Contributor • 1/4 Loading…
27
fix: rank ray.sub topo_rank 1..N instead of stripping SLURM_TOPOLOGY_ADDR digits community-request waiting-on-maintainers Waiting on maintainers to respond
#4102 opened Sep 11, 2026 by liayan Loading…
1
ProTip! Mix and match filters to narrow down what you’re looking for.