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

A complete Llama2 inference engine that fits in 1356 bytes of x86 assembly (github.com)

27 points by monax 1 day ago | discuss