Which best captures the key idea of "Top-p and sampling"?

ASpecific prompts remove guesswork - name the audience, length, tone, and constraints BTry the cheapest test that could prove you wrong CTop-p (nucleus) sampling limits choices to the most probable tokens, balancing coherence and variety DSometimes you only need to demonstrate the format, not the content