A Stitch in Time Saves Nine: Proactive Self-Refinement for Language Models
This addresses the inefficiency of reactive self-refinement methods for language model users, offering a novel approach to improve performance and reduce computational costs.
The paper tackles the problem of inefficient self-refinement in large language models by proposing ProActive Self-Refinement (PASR), which dynamically refines outputs during generation, resulting in a 41.6% reduction in token consumption and an 8.2% improvement in accuracy on Qwen3-8B.
Recent advances in self-refinement have demonstrated significant potential for improving the outputs of large language models (LLMs) through iterative refinement. However, most existing self-refinement methods rely on a reactive process with a fixed number of iterations, making it difficult to determine the optimal timing and content of refinement based on the evolving generation context. Inspired by the way humans dynamically refine their thoughts during execution, we propose ProActive Self-Refinement (PASR), a novel method that enables LLMs to refine their outputs during the generation process. Unlike methods that regenerate entire responses, PASR proactively decides whether, when, and how to refine based on the model's internal state and evolving context. We conduct extensive experiments on a diverse set of 10 tasks to evaluate the effectiveness of PASR. Experimental results show that PASR significantly enhances problem-solving performance. In particular, on Qwen3-8B, PASR reduces average token consumption by 41.6% compared to standard generation, while also achieving an 8.2% improvement in accuracy. Our code and baselines used in the paper are available on GitHub.