> LLMs don't have a concept of "best". Only most likely in what they've been trained on. At temperature 0 they are effectively producing the token that maximiz

> LLMs don't have a concept of "best". Only most likely in what they've been trained on. At temperature 0 they are effectively producing the token that maximizes a weighted sum of base LM probability and model reward.