Skip to content

Run frontier MoE models on hardware you already own — pure C, zero deps, experts streamed from disk. Tiny engine, immense model. 🐦

C

Commits on main

feat: add distributed expert workers
Gaurav Saini committed 4 months ago
0998991
ed75a25
36e8397
4d51576
eeaa6b6
12368b7
flake: build every supported engine, unpin COLI_ENGINE
Jérôme Benoit committed 3 months ago
0fd5228
refactor(core): make model families registry-owned
Mario Gutierrez committed 3 months ago
fe3879c
0578bba
85951ef
Merge pull request #1055 from dcutugno/pr/v4-cuda-tier
Vincenzo Fornaro committed 3 months ago
dd39c4e
03e8677
Merge pull request #934 from Nanetnounou/pr/neon-e8
Vincenzo Fornaro committed 3 months ago
3d5200d