this is actually faster transaltion than i imagined. i'd heard about groq, seems like they are better than nvidia in llms processing