Recursive Think-Answer Process for LLMs and VLMs

2026-03-02

AI Summary

This work addresses the vulnerability of large language models (LLMs) and vision-language models (VLMs) to output errors in single-pass inference, even when self-reflection cues are present, due to the absence of an effective error-correction mechanism. To overcome this limitation, the authors propose the Recursive Think-and-Answer Process (R-TAP), which employs an iterative reasoning loop coupled with a confidence assessment mechanism to iteratively refine model outputs. R-TAP incorporates a confidence generator and a dual-reward scheme—comprising recursive confidence-improvement rewards and final-answer confidence rewards—to substantially suppress unproductive self-reflection and enhance both reasoning stability and accuracy. Experimental results demonstrate that R-TAP consistently outperforms single-pass inference approaches across LLM and VLM tasks, achieving lower error rates, higher efficiency, and greater robustness.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Safety and Robustness
• Computer Vision: Large Vision Models

Application Category

• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Think-Answer reasoners such as DeepSeek-R1 have made notable progress by leveraging interpretable internal reasoning. However, despite the frequent presence of self-reflective cues like "Oops!", they remain vulnerable to output errors during single-pass inference. To address this limitation, we propose an efficient Recursive Think-Answer Process (R-TAP) that enables models to engage in iterative reasoning cycles and generate more accurate answers, going beyond conventional single-pass approaches. Central to this approach is a confidence generator that evaluates the certainty of model responses and guides subsequent improvements. By incorporating two complementary rewards-Recursively Confidence Increase Reward and Final Answer Confidence Reward-we show that R-TAP-enhanced models consistently outperform conventional single-pass methods for both large language models (LLMs) and vision-language models (VLMs). Moreover, by analyzing the frequency of "Oops"-like expressions in model responses, we find that R-TAP-applied models exhibit significantly fewer self-reflective patterns, resulting in more stable and faster inference-time reasoning. We hope R-TAP pave the way evolving into efficient and elaborated methods to refine the reasoning processes of future AI.

Problem

• Think-Answer reasoning
• output errors
• single-pass inference
• self-reflective cues
• reasoning stability

Innovation

• Recursive Think-Answer Process
• confidence-based reasoning
• iterative refinement
• large language models
• vision-language models