Really impressive demo! I didn't realize it was possible to run an LLM locally with such speed!