Stepwise Reasoning Checkpoint Analysis: A Test Time Scaling Method to Enhance LLMs' Reasoning

TL;DR

This paper introduces SRCA, a novel framework with checkpoints and clustering strategies that enhances the reasoning accuracy of Large Language Models by reducing path homogenization and utilizing intermediate results more effectively.

Contribution

The paper proposes Stepwise Reasoning Checkpoint Analysis (SRCA), a new method that improves test-time scaling for LLMs by introducing checkpoints and clustering to maintain diversity and leverage intermediate answers.

Findings

SRCA outperforms existing TTS methods in reasoning accuracy.

It effectively reduces path homogenization in reasoning paths.

The method demonstrates robustness and fault-tolerance in mathematical reasoning tasks.

Abstract

Mathematical reasoning through Chain-of-Thought (CoT) has emerged as a powerful capability of Large Language Models (LLMs), which can be further enhanced through Test-Time Scaling (TTS) methods like Beam Search and DVTS. However, these methods, despite improving accuracy by allocating more computational resources during inference, often suffer from path homogenization and inefficient use of intermediate results. To address these limitations, we propose Stepwise Reasoning Checkpoint Analysis (SRCA), a framework that introduces checkpoints between reasoning steps. It incorporates two key strategies: (1) Answer-Clustered Search, which groups reasoning paths by their intermediate checkpoint answers to maintain diversity while ensuring quality, and (2) Checkpoint Candidate Augmentation, which leverages all intermediate answers for final decision-making. Our approach effectively reduces path homogenization and creates a fault-tolerant mechanism by utilizing high-quality intermediate results. Experimental results show that SRCA improves reasoning accuracy compared to existing TTS methods across various mathematical datasets.