Putting everything in the prompt versus fine tunning my llm....is one faster than the other in terms of producing output