A 27B Multimodal Model Squeezed Into 5.9 Gigabytes
PrismML's ternary 27B multimodal model lands at 5.9GB with 98.2% of full-precision benchmark performance, and the retention number is the part worth checking.
local-ai
3 articles tagged “local-ai”.
A Pi 5 Frame Hears Birds and Draws 800 Hand-Cut 1800s Plates
A Raspberry Pi 5 listens for garden birds with BirdNET-Go and redraws an e-ink page using hand-cut 1800s illustrations, all running locally.
A 3-Star C Repo Gained 2,233 Stars in a Day
A pure-C engine with zero dependencies promises frontier MoE models on hardware you already own, and picked up 2,233 stars in a single day.