LLM inference is mostly about avoiding unnecessary movements btw

→ View original post on X — @theahmadosman