Research

Paper

AI LLM March 02, 2026

Recursive Think-Answer Process for LLMs and VLMs

Authors

Byung-Kwan Lee, Youngchae Chee, Yong Man Ro

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.

Metadata

arXiv ID: 2603.02099
Provider: ARXIV
Primary Category: cs.CL
Published: 2026-03-02
Fetched: 2026-03-03 04:34

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.02099v1</id>\n    <title>Recursive Think-Answer Process for LLMs and VLMs</title>\n    <updated>2026-03-02T17:20:10Z</updated>\n    <link href='https://arxiv.org/abs/2603.02099v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.02099v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>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.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CL'/>\n    <published>2026-03-02T17:20:10Z</published>\n    <arxiv:comment>CVPR 2026 Findings, Project page: https://litcoderr.github.io/rtap_page/</arxiv:comment>\n    <arxiv:primary_category term='cs.CL'/>\n    <author>\n      <name>Byung-Kwan Lee</name>\n    </author>\n    <author>\n      <name>Youngchae Chee</name>\n    </author>\n    <author>\n      <name>Yong Man Ro</name>\n    </author>\n  </entry>"
}