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

41b838a
09c4d96
570deac
Merge pull request #732 from michael-denyer/m5-metal-fmt6
Vincenzo Fornaro committed 3 months ago
618b5e9
Merge pull request #621 from jeswr/causal-ablation-harness
Vincenzo Fornaro committed 3 months ago
ba66ae5
Merge pull request #529 from monotophic/fmt7/stamp-registry
Vincenzo Fornaro committed 3 months ago
9458439
Merge pull request #620 from jeswr/docs-api-tools
Vincenzo Fornaro committed 3 months ago
6669ee9
3e6fdff
Merge pull request #716 from terrizoaguimor/route-telemetry
Vincenzo Fornaro committed 3 months ago
c5bfea7
Merge pull request #528 from monotophic/fmt7/fp8-passthrough
Vincenzo Fornaro committed 3 months ago
d2d6f6f
feat(kimi_k3): adopt the shared routing telemetry (#700)
Mario Gutiérrez committed 3 months ago
116f65a
b956472
ca3d1db
Merge pull request #673 from ZacharyZcR/feat/coli-tune
Vincenzo Fornaro committed 3 months ago
4b91559
b877e93
3d4f848
f90ca86
ea24cb0
Merge pull request #386 from monotophic/defaults/metal-cache
Vincenzo Fornaro committed 3 months ago
3bce2f0