Confidence Matters: Revisiting Intrinsic Self-Correction Capabilities of Large Language Models
It addresses the debate on self-correction feasibility in LLMs, offering a practical framework for researchers and practitioners to enhance model reliability, though it is incremental as it builds on existing prompting methods.
This paper investigates the self-correction capabilities of large language models (LLMs), identifying 'confidence' as a key latent factor that, when overlooked, leads to over-criticism and unreliable conclusions; it introduces an 'If-or-Else' (IoE) prompting framework that leverages this confidence to achieve consistent improvements in accuracy for self-corrected responses.
The recent success of Large Language Models (LLMs) has catalyzed an increasing interest in their self-correction capabilities. This paper presents a comprehensive investigation into the intrinsic self-correction of LLMs, attempting to address the ongoing debate about its feasibility. Our research has identified an important latent factor - the "confidence" of LLMs - during the self-correction process. Overlooking this factor may cause the models to over-criticize themselves, resulting in unreliable conclusions regarding the efficacy of self-correction. We have experimentally observed that LLMs possess the capability to understand the "confidence" in their own responses. It motivates us to develop an "If-or-Else" (IoE) prompting framework, designed to guide LLMs in assessing their own "confidence", facilitating intrinsic self-corrections. We conduct extensive experiments and demonstrate that our IoE-based Prompt can achieve a consistent improvement regarding the accuracy of self-corrected responses over the initial answers. Our study not only sheds light on the underlying factors affecting self-correction in LLMs, but also introduces a practical framework that utilizes the IoE prompting principle to efficiently improve self-correction capabilities with "confidence". The code is available at https://github.com/MBZUAI-CLeaR/IoE-Prompting.git.