-
Notifications
You must be signed in to change notification settings - Fork 407
Pull requests: NovaSky-AI/SkyRL
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
[tinker] 17/n towards Kimi K2.6: run Tinker API forward requests through the training loss path
#2068
opened Aug 19, 2026 by
casper-hansen
Contributor
Loading…
[tinker] 16/n towards Kimi K2.6: accept zstd-compressed Tinker API forward_backward bodies (SDK >= 0.25)
#2067
opened Aug 19, 2026 by
casper-hansen
Contributor
Loading…
[tinker] 15/n towards Kimi K2.6: continuous sampling for the Tinker API engine under colocate_all
#2066
opened Aug 19, 2026 by
casper-hansen
Contributor
Loading…
[tinker] 14/n towards Kimi K2.6: sleep colocated engines around Tinker API checkpoint ops
#2065
opened Aug 19, 2026 by
casper-hansen
Contributor
Loading…
[tinker] 13/n towards Kimi K2.6: adapter-only colocated weight sync for Tinker API LoRA (merge_lora=false)
#2064
opened Aug 19, 2026 by
casper-hansen
Contributor
Loading…
[megatron] 12/n towards Kimi K2.6: harden AdapterStore swaps for Tinker API multi-LoRA session churn
#2063
opened Aug 19, 2026 by
casper-hansen
Contributor
Loading…
[megatron] 11/n towards Kimi K2.6: file-backed offload of frozen masters so Tinker API LoRA training fits host RAM
#2062
opened Aug 19, 2026 by
casper-hansen
Contributor
Loading…
[Tinker API] Keep training payloads out of SQLite
#2061
opened Aug 19, 2026 by
yapdianang
Contributor
•
Draft
[Tinker API] Serialize SQLite training writes
#2060
opened Aug 19, 2026 by
yapdianang
Contributor
•
Draft
[Tinker API] Keep forwarded sample futures in memory
#2059
opened Aug 19, 2026 by
yapdianang
Contributor
•
Draft
[Tinker API] Cache validated sampling targets
#2058
opened Aug 19, 2026 by
yapdianang
Contributor
•
Draft
[feat][SFT] Add native Fireworks training backend
#2057
opened Aug 19, 2026 by
bharatmekala
Loading…
7 tasks done
fix(megatron): isolate Mamba state across packed sequences
#2056
opened Aug 19, 2026 by
dyurk-lila
Contributor
Loading…
Add Fireworks reserved capacity configuration
#2054
opened Aug 18, 2026 by
maanasbaraya
Loading…
4 tasks done
Fix hybrid FP8 KV-cache wake on vLLM 0.26
#2053
opened Aug 18, 2026 by
yapdianang
Contributor
Loading…
Run GSPO natively and harden Tinker transport
#2049
opened Aug 17, 2026 by
yapdianang
Contributor
•
Draft
[fix][tinker] Preserve training order and isolate sample futures
#2048
opened Aug 17, 2026 by
yapdianang
Contributor
•
Draft
[security][ci] Restrict privileged GPU jobs to trusted branches
#2047
opened Aug 16, 2026 by
bvolpato
Contributor
Loading…
[fix][train] Reject unsafe MTP prefix caching
#2046
opened Aug 16, 2026 by
bvolpato
Contributor
Loading…
[ci] Surface Anyscale logs for failed jobs
#2045
opened Aug 16, 2026 by
bvolpato
Contributor
Loading…
[Tinker API] Preserve input order for native GSPO updates
#2043
opened Aug 14, 2026 by
yapdianang
Contributor
Loading…
[deps][megatron] bump megatron-bridge to 0.7.0 and megatron-core to 0.20.0
run_h100_gpu_ci
Run H100 GPU CI
run_tinker_skyrl_train_backend_gpu_ci
run_train_megatron_gpu_ci_models
run_train_megatron_gpu_ci
#2042
opened Aug 14, 2026 by
erictang000
Collaborator
Loading…
[dependencies] Upgrade to cuda 13 by default
run_h100_gpu_ci
Run H100 GPU CI
#2040
opened Aug 14, 2026 by
erictang000
Collaborator
Loading…
1 task
Previous Next
ProTip!
Find all pull requests that aren't related to any open issues with -linked:issue.