Skip to content

Pull requests: NVIDIA-NeMo/RL

New pull request New
307 Open 1,857 Closed
307 Open 1,857 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

docs: fix stale references and typos for NVBug 6436630 CI:docs Run doctest Documentation Improvements or additions to documentation
#3147 opened Jul 9, 2026 by qiaochuz-nv Contributor Loading…
1
fix: ensure venv .dist-info entries have METADATA for downstream installs
#3146 opened Jul 9, 2026 by kajalj22 Contributor • Draft
2 tasks
3
fix: use staged SWE1 data in nightly recipe CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version)
#3145 opened Jul 9, 2026 by zyzhou5 Contributor Loading…
2
fix: router-replay startup validation and NeMo-Gym trace alignment CI:L1 Run doctests, unit tests, and functional tests
#3144 opened Jul 9, 2026 by zyzhou5 Contributor Loading…
3
ci: loosen val_loss threshold for sft test CI:Lfast Runs a fast test suite and re-use nightly `main` container (but sync dependencies to PRs version) r0.7.0
#3143 opened Jul 9, 2026 by ashors1 Contributor Loading…
4 tasks
2
cp: perf: perf recipe changes to enable force on policy ratio (3135) into r0.7.0 cherry-pick Run CICD
#3142 opened Jul 9, 2026 by svcnvidia-nemo-ci Contributor Loading…
2
feat(sc): add weight checkpoint save/restore to SingleController
#3138 opened Jul 9, 2026 by haitian-nvidia Loading…
1
build: bump vllm 0.20.0 → 0.24.0 CI:L1 Run doctests, unit tests, and functional tests
#3137 opened Jul 9, 2026 by kajalj22 Contributor • Draft
1 of 2 tasks
2
cp: fix: un-blocklist deepseek_v3 in transformers tokenizer class set (3116) into r0.7.0 cherry-pick CI:L1 Run doctests, unit tests, and functional tests
#3136 opened Jul 9, 2026 by svcnvidia-nemo-ci Contributor Loading…
2
ci: Bump Megatron-Bridge to 1523d2f CI:L1 Run doctests, unit tests, and functional tests
#3134 opened Jul 9, 2026 by svcnvidia-nemo-ci Contributor Loading…
3
docs: add review-pr-team contributor skill for multi-agent PR review
#3133 opened Jul 9, 2026 by terrykong Collaborator Loading…
2 of 4 tasks
1
feat: add OpenR1-Math-220k response dataset community-request Documentation Improvements or additions to documentation
#3131 opened Jul 9, 2026 by tianyi-zhang-02 Contributor Loading…
1
[WIP] SWE agentic training for trtllm rollout
#3130 opened Jul 8, 2026 by hchings • Draft
4 tasks
1
feat: router replay unified wrapper CI:L1 Run doctests, unit tests, and functional tests
#3129 opened Jul 8, 2026 by zyzhou5 Contributor Loading…
5
feat: support fused selected-token logprobs for AutoModel community-request Documentation Improvements or additions to documentation
#3128 opened Jul 8, 2026 by mohamed-alansary Loading…
1
feat(sglang): enable NeMo-Gym async-GRPO rollouts on the SGLang backend
#3127 opened Jul 8, 2026 by Kh4L • Draft
1
fix: rebuild async GRPO dataloaders inside Ray collectors community-request
#3125 opened Jul 8, 2026 by mohamed-alansary Loading…
1
perf: reduce selected-token logprob memory community-request
#3124 opened Jul 8, 2026 by mohamed-alansary Loading…
1
fix: skip no-op TP collectives for world size one community-request
#3123 opened Jul 8, 2026 by mohamed-alansary Loading…
1
fix: normalize packed broadcast tensors onto CUDA before NCCL broadcast community-request
#3122 opened Jul 8, 2026 by mohamed-alansary Loading…
1
fix: complete truncated html_theme_options in conf.py (closes #2816) community-request Documentation Improvements or additions to documentation
#3121 opened Jul 8, 2026 by botbikamordehai2-sketch Loading…
1
1
feat: size routed_experts carry dtype from the model's expert count CI:L1 Run doctests, unit tests, and functional tests
#3113 opened Jul 8, 2026 by zyzhou5 Contributor Loading…
5
docs: add cmake and ninja-build as bare-metal prerequisites Documentation Improvements or additions to documentation
#3111 opened Jul 8, 2026 by kajalj22 Contributor • Draft
2 tasks
1
2
chore: replace decord2 with torchvision/torchaudio for media loading CI:L1 Run doctests, unit tests, and functional tests r0.7.0
#3107 opened Jul 7, 2026 by kajalj22 Contributor Loading…
5
Support FFN refit aliases for DeepSeek V4, Qwen3.5, and GLM Documentation Improvements or additions to documentation
#3104 opened Jul 7, 2026 by youngeunkwon0405 Contributor • Draft
1
ProTip! Add no:assignee to see everything that’s not assigned.