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

(github.com) by monax | May 5, 2026 | 0 comments on HN