Announcement_8

Our paper on a ‘Speculate-and-Refine’ async streaming framework for on-device LLM inference has been accepted to MobiCom 2026.