A complete Llama2 inference engine that fits in 1356 bytes of x86 assembly

 (github.com)

14 points | by monax 8 hours ago

0 comments