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

32d513d
Merge pull request #1492 from eltociear/add-ja-doc
Vincenzo Fornaro committed 2 months ago
3fa3162
Merge pull request #1606 from heyyyyjude/xeon-k1b-tile
Vincenzo Fornaro committed 2 months ago
1d85905
dc3b879
201af1c
3dca747
Merge pull request #1239 from jtinbergen/sse41-tier-upstream
Vincenzo Fornaro committed 2 months ago
3827f42
Merge pull request #1313 from BrianHeeseIs/pr/fp8-4row
Vincenzo Fornaro committed 2 months ago
5d070b0
7ec9f09
87e7153
f54ba99
Merge pull request #1610 from JohnRTitor/fix-build
Vincenzo Fornaro committed 2 months ago
cac9a0e
a4ce3a5
0e724f1
Merge pull request #1639 from JustVugg/docs/readme-brio
Vincenzo Fornaro committed 2 months ago
2a88335
099aac9
Merge pull request #1638 from JustVugg/dev
Vincenzo Fornaro committed 2 months ago
dcd7383
test: let the long-path case run only where the path fits PATH_MAX
Marcello Costagliola committed 2 months ago
9ec0f55
Merge pull request #1636 from JustVugg/release/1.12.0
Vincenzo Fornaro committed 2 months ago
2fe3b90
84dda8b
changelog: the three forms of /v1/brio
JustVugg committed 2 months ago
37c54e1