I don't understand what you mean: I have no clue about anything related to reverse engineering, but I ported the mistral tokenizer to Rust and also wrote a basic CPU Llama training and inference implementation in Rust, so I definitely wouldn't need an intro to model inference…