What does it mean that an LLM is a reward model ? , I always thought of rewards only in the RL field . And how would the reward model be used during finetuning?
What does it mean that an LLM is a reward model ? , I always thought of rewards only in the RL field . And how would the reward model be used during finetuning?