Learning to Check: Unleashing Potentials for Self-Correction in Large Language Models

TL;DR

This paper introduces a new training format called 'Step CoT Check' to improve large language models' ability to self-detect and correct reasoning errors, significantly enhancing their self-checking performance across benchmarks.

Contribution

It proposes a novel 'Step CoT Check' format and dataset for training LLMs, leading to improved self-correction abilities, especially in reasoning tasks.

Findings

Enhanced error detection and correction in LLMs.

Significant performance gains on multiple benchmarks.

Greater benefits observed in larger models.

Abstract

Self-correction has achieved impressive results in enhancing the style and security of the generated output from large language models (LLMs). However, recent studies suggest that self-correction might be limited or even counterproductive in reasoning tasks due to LLMs' difficulties in identifying logical mistakes. In this paper, we aim to enhance the self-checking capabilities of LLMs by constructing training data for checking tasks. Specifically, we apply the Chain of Thought (CoT) methodology to self-checking tasks, utilizing fine-grained step-level analyses and explanations to assess the correctness of reasoning paths. We propose a specialized checking format called "Step CoT Check". Following this format, we construct a checking-correction dataset that includes detailed step-by-step analysis and checking. Then we fine-tune LLMs to enhance their error detection and correction abilities. Our experiments demonstrate that fine-tuning with the "Step CoT Check" format significantly improves the self-checking and self-correction abilities of LLMs across multiple benchmarks. This approach outperforms other formats, especially in locating the incorrect position, with greater benefits observed in larger models. For reproducibility, all the datasets and code are provided in https://github.com/bammt/Learn-to-check.