Process Reward Model
An advanced reward model used in RLHF that evaluates and scores each individual step of a model's reasoning chain, rather than just the final answer.
Think of It Like This
Like a math teacher giving a student partial credit for correctly executing the first three steps of a formula, even if they messed up the final calculation.
Traditional outcome reward models (ORM) only verify if the final output is correct. PRMs provide dense, fine-grained feedback, explicitly rewarding logical consistency at every step. This drastically reduces hallucinations in complex reasoning tasks like mathematics or coding, as the model learns to favor rigorous methodologies over lucky guesses.