Beyond the First Error: Process Reward Models for Reflective Mathematical Reasoning
read the original abstract
Many studies focus on data annotation techniques for training effective PRMs. However, current methods encounter a significant issue when applied to long CoT reasoning processes: they tend to focus solely on the first incorrect step and all preceding steps, assuming that all subsequent steps are incorrect. These methods overlook the unique self-correction and reflection mechanisms inherent in long CoT, where correct reasoning steps may still occur after initial reasoning mistakes. To address this issue, we propose a novel data annotation method for PRMs specifically designed to score the long CoT reasoning process. Given that under the reflection pattern, correct and incorrect steps often alternate, we introduce the concepts of Error Propagation and Error Cessation, enhancing PRMs' ability to identify both effective self-correction behaviors and reasoning based on erroneous steps. Leveraging an LLM-based judger for annotation, we collect 1.7 million data samples to train a 7B PRM and evaluate it at both solution and step levels. Experimental results demonstrate that compared to existing open-source PRMs and PRMs trained on open-source datasets, our PRM achieves superior performance across various metrics, including search guidance, BoN, and F1 scores. Compared to widely used MC-based annotation methods, our annotation approach not only achieves higher data efficiency but also delivers superior performance. Detailed analysis is also conducted to demonstrate the stability and generalizability of our method.
This paper has not been read by Pith yet.
Forward citations
Cited by 5 Pith papers
-
PDCR: Perception-Decomposed Confidence Reward for Vision-Language Reasoning
PDCR improves vision-language reasoning by computing separate normalized confidence advantages for perception steps and reasoning steps after unsupervised decomposition.
-
Internalizing Outcome Supervision into Process Supervision: A New Paradigm for Reinforcement Learning for Reasoning
A new RL paradigm for reasoning where models generate their own internal process supervision from outcome feedback by recycling failed trajectories.
-
Improving Medical VQA through Trajectory-Aware Process Supervision
A trajectory-aware process reward using DTW on sentence embeddings, combined with exact-match in GRPO after SFT, raises mean medical VQA accuracy from 0.598 to 0.689 across six benchmarks.
-
Internalizing Outcome Supervision into Process Supervision: A New Paradigm for Reinforcement Learning for Reasoning
A method internalizes outcome supervision into process supervision by extracting step-level learning signals from failed reasoning trajectories during reinforcement learning.
-
LLM Reasoning with Process Rewards for Outcome-Guided Steps
PROGRS uses outcome-conditioned centering on PRM scores to safely integrate process rewards into GRPO for improved Pass@1 on math benchmarks.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.