Skip to content

Pull requests: NVIDIA-NeMo/RL

New pull request New
296 Open 1,753 Closed
296 Open 1,753 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

ci: Bump Megatron-Bridge to 97e8dba CI:L1 Run doctests, unit tests, and functional tests
#3025 opened Jul 1, 2026 by svcnvidia-nemo-ci Contributor Loading…
3
cp: feat: R3 gym notq router replay (2915) into r0.7.0 cherry-pick CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version)
#3023 opened Jul 1, 2026 by svcnvidia-nemo-ci Contributor Loading…
2
fix: (qwen3-omni) disable vLLM sequence_parallel_chunk custom op r0.7.0
#3021 opened Jul 1, 2026 by yuekaizhang Contributor Loading…
1 task
1
cp: feat(xtoken): multi-teacher support for cross-tokenizer off-policy distillation (2797) into r0.7.0 cherry-pick CI:L1 Run doctests, unit tests, and functional tests Documentation Improvements or additions to documentation
#3019 opened Jul 1, 2026 by svcnvidia-nemo-ci Contributor Loading…
2
fix: Keep QARL nightlies schedulable on H100 nodes CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version) r0.7.0
#3018 opened Jul 1, 2026 by mxinO Contributor Loading…
4 tasks
9
ci: fix grpo-nanov3-30BA3B-2n8g-megatron_generation.yaml config CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version) r0.7.0
#3017 opened Jul 1, 2026 by ashors1 Contributor Loading…
4 tasks
2
ci: Add nightly GB200 super test CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version)
#3016 opened Jun 30, 2026 by ashors1 Contributor Loading…
4 tasks
3
feat: Llama true-on-policy mode
#3015 opened Jun 30, 2026 by guyueh1 Contributor • Draft
4 tasks
1
fix: reduce NUM_MINUTES to 240 in dapo-deepseek-v3-64n8g.v2 test
#3014 opened Jun 30, 2026 by kajalj22 Contributor Loading…
1 task
1
fix: stabilize GB200 16n4g issue-2579 repro CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version)
#3013 opened Jun 30, 2026 by jinglinglingling Contributor Loading…
1
2
Mxin/simulated kv cache qarl Documentation Improvements or additions to documentation
#3012 opened Jun 30, 2026 by mxinO Contributor • Draft
4 tasks
2
fix(perf): fix CUDA OOM in grpo-qwen3-30ba3b-4n8g performance test CI:L1 Run doctests, unit tests, and functional tests r0.7.0
#3005 opened Jun 30, 2026 by NolenLiang Contributor Loading…
1
4
feat(grpo): stream rollout batches by prompt group 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
#3000 opened Jun 30, 2026 by yfw Contributor Loading…
4 tasks
4
feat(vllm): MTP speculative-decoding inference with refit drafter
#2999 opened Jun 30, 2026 by yfw Contributor Loading…
4 tasks
1
refactor(sglang): allocate ports from reserved ranges community-request waiting-on-customer Waiting on the original author to respond
#2997 opened Jun 29, 2026 by xiuhu17 Contributor Loading…
1
9
Update super recipes for gym bump
#2996 opened Jun 29, 2026 by yfw Contributor Loading…
4 tasks
3
feat: Dynamo K8s Integration Documentation Improvements or additions to documentation
#2990 opened Jun 29, 2026 by jthomson04 Contributor Loading…
4 tasks
5
fix(megatron): prefer nvrx strategy for HF import saves CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version) r0.7.0
#2989 opened Jun 29, 2026 by jinglinglingling Contributor Loading…
6
feat(modelopt): support real NVFP4 QAT rollout for MoE and Mamba CI:L1 Run doctests, unit tests, and functional tests Documentation Improvements or additions to documentation Feature
#2983 opened Jun 29, 2026 by HollowMan6 Member • Draft
4 tasks done
1
15
refactor(ppo): reuse LossPostProcessor in value worker + add offload_to_cpu CI:L1 Run doctests, unit tests, and functional tests
#2980 opened Jun 28, 2026 by bg51717 Contributor Loading…
3 of 4 tasks
1
5
fix: complete truncation in docs/conf.py html_theme_options (closes #2816) community-request Documentation Improvements or additions to documentation waiting-on-maintainers Waiting on maintainers to respond
#2979 opened Jun 28, 2026 by botbikamordehai2-sketch Loading…
1
1
Use JSON for vLLM env list parsing community-request waiting-on-maintainers Waiting on maintainers to respond
#2978 opened Jun 28, 2026 by fallintoplace Loading…
3 of 4 tasks
1
Avoid leaking env vars across Ray worker groups community-request waiting-on-maintainers Waiting on maintainers to respond
#2977 opened Jun 28, 2026 by fallintoplace Loading…
3 of 4 tasks
1
Fix setuptools package discovery for nemo_rl subpackages community-request waiting-on-maintainers Waiting on maintainers to respond
#2976 opened Jun 28, 2026 by fallintoplace Loading…
3 of 4 tasks
1
Add Nano v3 async GRPO config
#2973 opened Jun 28, 2026 by snowmanwwg Contributor Loading…
1
ProTip! no:milestone will show everything without a milestone.