Keep the LoRA runtime warm between jobs - #1997

Draft

j316chuck wants to merge 4 commits into

NovaSky-AI:mainfrom

j316chuck:chuck-codex/keep-runtime-warm-live

Commits on Jul 23, 2026

Prewarm Tinker inference before model creation
j316chuck
committed
52359dc
Keep prewarmed inference alive after model unload
j316chuck
committed
1d1758f

Commits on Aug 6, 2026

Add option to retain warm LoRA runtime after unload
j316chuck
committed
1eaf961
Fix prewarm engine test backend mock
j316chuck
committed
ddf7b24