GenPRM: Scaling Test-Time Compute of Process Reward Models via Generative Reasoning
This work addresses the problem of scaling test-time compute and improving generalization for process supervision in LLMs, establishing a new paradigm that bridges PRMs and critic models.
The paper tackles the limitations of Process Reward Models (PRMs) in large language models by introducing GenPRM, which uses generative reasoning with code verification to improve process supervision; it achieves significant performance gains, such as a 1.5B model outperforming GPT-4o and a 7B model surpassing Qwen2.5-Math-PRM-72B on ProcessBench with only 23K training data.
Recent advancements in Large Language Models (LLMs) have shown that it is promising to utilize Process Reward Models (PRMs) as verifiers to enhance the performance of LLMs. However, current PRMs face three key challenges: (1) limited process supervision and generalization capabilities, (2) dependence on scalar value prediction without leveraging the generative abilities of LLMs, and (3) inability to scale the test-time compute of PRMs. In this work, we introduce GenPRM, a generative process reward model that performs explicit Chain-of-Thought (CoT) reasoning with code verification before providing judgment for each reasoning step. To obtain high-quality process supervision labels and rationale data, we propose Relative Progress Estimation (RPE) and a rationale synthesis framework that incorporates code verification. Experimental results on ProcessBench and several mathematical reasoning tasks show that GenPRM significantly outperforms prior PRMs with only 23K training data from MATH dataset. Through test-time scaling, a 1.5B GenPRM outperforms GPT-4o, and a 7B GenPRM surpasses Qwen2.5-Math-PRM-72B on ProcessBench. Additionally, GenPRM demonstrates strong abilities to serve as a critic model for policy model refinement. This work establishes a new paradigm for process supervision that bridges the gap between PRMs and critic models in LLMs. Our code, model, and data will be available in https://ryanliu112.github.io/GenPRM.