Skip to content

Pull requests: PrimeIntellect-ai/prime-rl

New pull request New
132 Open 2,523 Closed
132 Open 2,523 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

chore: migrate off v0 logic-env/math-env to v1 tasksets
#2999 opened Jul 11, 2026 by mikasenghaas Member • Draft
2
chore: release v0.7.0
#2998 opened Jul 10, 2026 by samsja Member • Draft
exp(glm45air): rubric monitoring/shaping configs (rlm + swe judges, research/glm-5.2)
#2997 opened Jul 10, 2026 by mikasenghaas Member • Draft
exp(glm45air): loss ablation configs (dppo / ipo / kimi-k1.5 scaffold)
#2996 opened Jul 10, 2026 by mikasenghaas Member • Draft
feat(orchestrator): support infinite tasksets
#2993 opened Jul 10, 2026 by mikasenghaas Member • Draft
feat(sft): require renderer-based tokenization
#2988 opened Jul 10, 2026 by hubert-marek Contributor Loading…
24
fix(orchestrator): propagate dispatcher and watcher failures
#2986 opened Jul 9, 2026 by morluto Loading…
fix(orchestrator): use raw reward for singleton MaxRL groups
#2985 opened Jul 9, 2026 by morluto Loading…
1
fix: preserve explicit None config overrides across config write/re-parse
#2981 opened Jul 9, 2026 by mikasenghaas Member • Draft
feat: top-p/top-k train sampling with kept-set mask replay
#2979 opened Jul 9, 2026 by faresobeid Contributor • Draft
fix: bound RL importance ratios to prevent non-finite gradients
#2973 opened Jul 7, 2026 by morluto Loading…
1
7
Feat/rl check preflight
#2970 opened Jul 7, 2026 by anravich13-cloud Contributor Loading…
8
feat: TOP-D algorithm (trust region policy distillation)
#2966 opened Jul 7, 2026 by hallerite Member • Draft
feat(dynamo): add DynamoGraphDeployment Helm deployment for k8s - stacked PR (#2964)
#2965 opened Jul 6, 2026 by biswapanda Loading…
3
feat(dynamo): add native inference backend
#2964 opened Jul 6, 2026 by biswapanda Loading…
6
chore: bump verifiers to assume pre-installed v1 environments
#2963 opened Jul 6, 2026 by mikasenghaas Member • Draft
feat: train on replayed rollouts — continue (compaction/tool-call) + recheck envs
#2962 opened Jul 6, 2026 by faresobeid Contributor • Draft
fix(slurm): tear down inference-only nodes after clean trainer exit
#2961 opened Jul 5, 2026 by eligotts Contributor • Draft
chore: add environment deps auto-bump workflow
#2958 opened Jul 4, 2026 by samsja Member • Draft
feat: static SFT through the orchestrator (train on a HF dataset, no inference needed)
#2954 opened Jul 3, 2026 by tim0120 Contributor • Draft
feat(trainer): MoE loss-free load balancing and block dropout
#2953 opened Jul 3, 2026 by faresobeid Contributor Loading…
11
feat(trainer): fold rl/ref_kl loss types into one pg skeleton, vectorize compute_loss
#2951 opened Jul 3, 2026 by hallerite Member • Draft
feat(inference): HiSparse vLLM wheel for GLM-5.x PD decode instances
#2949 opened Jul 3, 2026 by faresobeid Contributor • Draft
feat(rl): pack multimodal samples with context parallelism
#2947 opened Jul 3, 2026 by hubert-marek Contributor Loading…
2 tasks done
5
feat(cp): context parallelism for hybrid + VLM SFT
#2946 opened Jul 3, 2026 by hubert-marek Contributor Loading…
2 tasks done
7
ProTip! Type g p on any issue or pull request to go back to the pull request listing page.