Skip to content

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

C

Commits on main

998c2f9
e3a0e78
Merge pull request #690 from terrizoaguimor/corpus-drafts
Vincenzo Fornaro committed 3 months ago
52dadb7
720e280
3f85dfa
fix(cuda): apply grouped scales in ragged attention
ZacharyZcR committed 3 months ago
67a07fd
da7e972
Merge pull request #681 from MrMetric/fix/hip-non-wmma-archs
Vincenzo Fornaro committed 3 months ago
0e825da
c67dbab
a3a5a75
659dbd6
96fd82b
c616ea6
Merge pull request #677 from JustVugg/dev
Vincenzo Fornaro committed 3 months ago
2d62381