Skip to content

Pull requests: NovaSky-AI/SkyRL

New pull request New
176 Open 1,328 Closed
176 Open 1,328 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

[feat] top-p sampler replay
#1901 opened Jul 15, 2026 by erictang000 Collaborator Loading…
3
[logging] Rollout time split + async-trainer GPU/phase/buffer observability
#1900 opened Jul 14, 2026 by eicherseiji Contributor • Draft
FP8 KV-Cache Compatibility for Hybrid vLLM Models
#1899 opened Jul 14, 2026 by YJHMITWEB Loading…
@jinghanyao1-hub
2
FP8 RL Training and Rollout Weight Synchronization
#1898 opened Jul 14, 2026 by YJHMITWEB Loading…
@jinghanyao1-hub
3
[algorithm] Add REPO-R adaptive entropy loss
#1896 opened Jul 14, 2026 by erictang000 Collaborator Loading…
5
[deps] Make transformers/ml_dtypes overrides unconditional to fix macOS dev env
#1894 opened Jul 14, 2026 by kurtislin Contributor Loading…
5 of 6 tasks
[fix] Fix training and evaluation loss metric calculation in SFT codepath
#1893 opened Jul 14, 2026 by SumanthRH Member • Draft
1 task
Add AMD ROCm Docker image and Tinker training examples
#1892 opened Jul 14, 2026 by eddierichter-amd Loading…
15
Add bounded-memory training and colocation controls
#1890 opened Jul 13, 2026 by YJHMITWEB Loading…
@jinghanyao1-hub
3
[deps] bump megatron-bridge to fbf2d3 run_h100_gpu_ci Run H100 GPU CI run_train_megatron_gpu_ci_models run_train_megatron_gpu_ci
#1888 opened Jul 13, 2026 by erictang000 Collaborator Loading…
[bugfix] Fused LM Head Compatibility with HybridModel
#1886 opened Jul 13, 2026 by dyurk-lila Contributor Loading…
@erictang000
4
[megatron] Enable grouped outer expert LoRA targeting for higher accuracy
#1884 opened Jul 11, 2026 by casper-hansen Contributor Loading…
6
[feat][SFT] Support multiple datasets for training and evaluation in SFTTrainer
#1883 opened Jul 10, 2026 by avigyabb Collaborator • Draft
1
1
[fix] Return loss_fn_outputs in input order with token-based microbatching
#1882 opened Jul 10, 2026 by erictang000 Collaborator Loading…
2
integrations/arctic_rl: Harbor recipe via override_entrypoint dispatch (+ design doc)
#1879 opened Jul 10, 2026 by sfc-gh-kganesan Contributor • Draft
4
[fix] Avoid inference engine initialization for the SFT code path
#1878 opened Jul 9, 2026 by avigyabb Collaborator Loading…
6
[example] add Fireworks Eval Only
#1871 opened Jul 7, 2026 by kyuds Contributor Loading…
2
[fsdp] Exclude fully-padding microbatches from metric aggregation (parity with #1817)
#1863 opened Jul 3, 2026 by kurtislin Contributor Loading…
5
[chore] Remove weight-sync hacks now handled natively by vLLM 0.23.0
#1861 opened Jul 1, 2026 by hao-aaron Collaborator Loading…
6
feat(algo): add cispo_anchor to select old vs rollout IS-ratio anchor
#1859 opened Jul 1, 2026 by tbalestri-lila • Draft
[chore] bump vllm to 0.25.0 run_train_gpu_ci run_train_megatron_gpu_ci_models run_train_megatron_gpu_ci
#1854 opened Jul 1, 2026 by erictang000 Collaborator Loading…
1
[env][examples] Add tau-bench (retail) environment + eval
#1853 opened Jun 30, 2026 by xyuzh • Draft
[tinker] Remove backend-specific config leak in sample batching
#1851 opened Jun 30, 2026 by kurtislin Contributor Loading…
[tinker] Retry external sample requests through the lazy adapter-load race
#1847 opened Jun 29, 2026 by john-hahn Contributor Loading…
4 tasks done
5
[codex] Add Tinker checkpoint delete endpoint
#1840 opened Jun 26, 2026 by j316chuck Loading…
1
@pcmoritz
4
ProTip! Type g p on any issue or pull request to go back to the pull request listing page.