Run a 1-billion parameter LLM on a $10 board with 256MB RAM - View it on GitHub