JustVugg/colibri

6 stories

JustVugg/colibri is a pure C inference engine for running frontier MoE models locally on hardware users already own. Recent GitHub Trending listings describe it as a tiny engine that streams expert weights from disk, aiming to lower memory and hardware requirements for local MoE inference.

JustVugg/colibri

Tiny C engine to run frontier MoE models, experts streamed from disk

▲ 98 · GitHub Trending ·
That is everything