Comment by amelius

Maybe I don't understand the usecase but I'm curious why speed matters given that LLMs are so slow (?)