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

release: bump version to 1.2.0
JustVugg committed 3 months ago
b0dbfdf
docs: add Discord community link
ZacharyZcR committed 3 months ago
00400e2
docs: publish open inference experiments
ZacharyZcR committed 3 months ago
4de7f4c
docs: highlight inference engine techniques
ZacharyZcR committed 3 months ago
04bdc86
docs: align translated project positioning
ZacharyZcR committed 3 months ago
e42541c
docs: remove the minimum-RAM headline
ZacharyZcR committed 3 months ago
22b505a
docs: define the inference research mission
ZacharyZcR committed 3 months ago
b583e9c
0877d8d
83c6f08
2a301ea
Merge pull request #655 from steve-m/feat-e8-converter
Vincenzo Fornaro committed 3 months ago
fa599e0
Merge pull request #654 from steve-m/feat-e8-avx2
Vincenzo Fornaro committed 3 months ago
346f4f1
quant: NEON rows16 fp4 kernels for aarch64
andres-gb10 committed 4 months ago
4aaf564
build: enable DeepSeek V4 engine on aarch64 Linux
andres-gb10 committed 4 months ago
4eadd93