but how would you run a local llm on a phone?
Very small models, good for stuff like offline translations and stuff, you do it with llama.cpp probably