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

 (github.com)

26 points | by monax a day ago ago

No comments yet.