Confidence v.s. Critique: A Decomposition of Self-Correction Capability for LLMs
This work addresses the issue of declining accuracy after self-correction in LLMs, which is important for developers and users seeking reliable AI systems, though it is incremental as it builds on existing self-correction research.
The paper tackles the problem of understanding and improving self-correction in Large Language Models by decomposing it into confidence and critique capabilities, proposing metrics to measure them, and finding that a strategy based on transforming Supervision Fine-Tuning data outperforms vanilla SFT, achieving higher accuracy after self-correction.
Large Language Models (LLMs) can correct their self-generated responses, but a decline in accuracy after self-correction is also witnessed. To have a deeper understanding of self-correction, we endeavor to decompose, evaluate, and analyze the self-correction behaviors of LLMs. By enumerating and analyzing answer correctness before and after self-correction, we decompose the self-correction capability into confidence (being confident to correct answers) and critique (turning wrong answers to correct) capabilities, and propose two metrics from a probabilistic perspective to measure these 2 capabilities, along with another metric for overall self-correction capability evaluation. Based on our decomposition and evaluation metrics, we conduct extensive experiments and draw some empirical conclusions. For example, we find different models can exhibit distinct behaviors: some models are confident while others are more critical. We also find the trade-off between the two capabilities (i.e. improving one can lead to a decline in the other) when manipulating model self-correction behavior by prompts or in-context learning. Further, we find a simple yet efficient strategy to improve self-correction capability by transforming Supervision Fine-Tuning (SFT) data format, and our strategy outperforms vanilla SFT in both capabilities and achieves much higher accuracy after self-correction. Our code will be publicly available on GitHub.