Can LLMs Learn from Previous Mistakes? Investigating LLMs' Errors to Boost for Reasoning

Anonymous

Abstract:

Large language models~(LLMs) have demonstrated striking reasoning capability. Recent works have shown the benefits to LLMs from fine-tuning golden-standard CoT rationales or using them as correct examples in few-shot prompting. While humans can indeed imitate correct examples, learning from our mistakes is another vital aspect of human cognition. Hence, a question naturally arises: \textit{can LLMs learn and benefit from their mistakes, especially for their reasoning? } This study investigates this problem from both the prompting and model-tuning perspectives. We begin by introducing \textsc{CoTErrorSet}, a new benchmark with 609,432 questions, each designed with both correct and error references, and demonstrating the types and reasons for making such mistakes. To explore the effectiveness of those mistakes, we design two methods: (1) Self-rethinking prompting guides LLMs to rethink whether they have made similar previous mistakes;
and (2) Mistake tuning involves finetuning models in both correct and incorrect reasoning domains, rather than only tuning models to learn ground truth in traditional methodology. We conduct a series of experiments to prove LLMs can obtain benefits from mistakes.
Both of our two methods serve as potential low-cost solutions to utilize mistakes to improve reasoning abilities compared with the high cost of making hand-crafted references. We ultimately make a thorough analysis of the reasons behind LLMs' errors, which provides directions that future research needs to overcome. \textsc{CoTErrorSet} will be published soon on \texttt{Anonymity Link}.

Paper Type: long

Research Area: Resources and Evaluation

Contribution Types: Model analysis & interpretability, NLP engineering experiment, Approaches to low-resource settings, Approaches low compute settings-efficiency, Data resources, Surveys

Languages Studied: English