Hardware independent LLM inference engine from ZML.

→ View original post on X — @ylecun