Three methods to give AI feedback when answers are hard to verify

Original: “Three ways to the unverifiable: Rubrics as Rewards, Generative Reward Models, Process Rewards”

Source: arxiv.org ↗