Skip to content

Pull requests: NovaSky-AI/SkyRL

New pull request New
86 Open 870 Closed
86 Open 870 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] Fix failing docs build; Fix GPU OOM for new inference codepath
#1221 opened Feb 26, 2026 by SumanthRH • Draft
[train] Remove deprecated SGLang backend from skyrl/
#1220 opened Feb 26, 2026 by SumanthRH • Draft
Bump vLLM to 0.16.0 and transformers to >=5.0.0 for GLM-4.7-Flash
#1218 opened Feb 25, 2026 by tyler-griggs • Draft
4
Add GLM-4.7-Flash (30B MoE) training example
#1215 opened Feb 25, 2026 by tyler-griggs • Draft
6
Register GLM-4.7-Flash bridge and bump megatron-bridge
#1214 opened Feb 25, 2026 by tyler-griggs • Draft
2
Add configurable MoE runtime flags to MegatronConfig
#1213 opened Feb 25, 2026 by tyler-griggs • Draft
4
Fix Megatron backend correctness: grad_scale_func, PP seed variation, weight sync pause
#1212 opened Feb 25, 2026 by tyler-griggs • Draft
6
[train][tests] Add tests for RemoteInferenceClient
#1211 opened Feb 25, 2026 by SumanthRH Loading…
60
Add llm_as_a_judge_local example with frozen vLLM reward model
#1208 opened Feb 25, 2026 by ghShu Loading…
22
[tx] Implement qwen3.5 model architecture
#1201 opened Feb 24, 2026 by pcmoritz • Draft
[tx] Add Qwen3 VL model tx
#1196 opened Feb 23, 2026 by tamoghnokandar • Draft
9
Deepcopy env_extras dicts for each sample when preparing generator input
#1189 opened Feb 20, 2026 by ebronstein Loading…
4
Use the last LoRA path in the vLLM inference engine instead of "dummy_lora_path"
#1188 opened Feb 20, 2026 by ebronstein Loading…
5
[tx] Fix engine command on TPUs tx
#1181 opened Feb 19, 2026 by pcmoritz Loading…
8
Restrict mini-swe-agent version to <2.0
#1180 opened Feb 19, 2026 by bluerose73 Loading…
5
[WIP][AsyncRL] Use keep mode for pause and resume
#1179 opened Feb 19, 2026 by hao-aaron • Draft
2 of 3 tasks
[Draft] support for azure blob storage
#1172 opened Feb 18, 2026 by chiragbhatt311 Loading…
17
Change asyncio.gather to TaskGroup in the Harbor generator to correctly propagate cancellation signals.
#1160 opened Feb 16, 2026 by alexgshaw • Draft
27
[tx] Implement context parallelism in tx with ring attention using ppermute tx
#1149 opened Feb 16, 2026 by tanmaysachan • Draft
1 task done
20
Add MaxRL mean normalization over advantages
#1126 opened Feb 15, 2026 by tamoghnokandar Loading…
1
8
[WIP] [tx] Add deterministic mode tx
#1125 opened Feb 15, 2026 by pcmoritz Loading…
3
[docs] Add contributions welcome banner to limitations page
#1119 opened Feb 13, 2026 by tyler-griggs • Draft
Support cloud model paths (s3://, gs://) for fresh training runs
#1113 opened Feb 13, 2026 by bthecohen • Draft
9
R3 PR: Rollout Routing Replay
#1101 opened Feb 13, 2026 by devpatelio Loading…
@CharlieFRuan
10
Add support for Markovian Text Environments
#1094 opened Feb 12, 2026 by DhananjayAshok Loading…
5
ProTip! Adding no:label will show everything without a label.