JustVugg/colibri

A hierarchical inference engine for frontier MoE models: hot experts on the GPU (CUDA, Vulkan), warm ones in RAM, cold ones streamed from NVMe. Written in C, runs on the hardware you already own.

★ 48k · 5.6k forks · C

View on GitHub

Add it to your README

Charts use snapshots recorded by Repostream, starting at the first observation. README images are cached for 6 to 12 hours. Limited history is shown as collecting, never estimated.

Star history chart

[![Star history](https://repostream.com/chart/JustVugg/colibri.svg)](https://repostream.com/repo/JustVugg/colibri)

Trending rank badge

[![Trending on Repostream](https://repostream.com/badge/JustVugg/colibri.svg)](https://repostream.com/repo/JustVugg/colibri)

Want to see what is gaining stars fastest right now? Open the live leaderboard.