何时思考，何时回顾：不确定性引导的回顾
When to Think and When to Look: Uncertainty-Guided Lookback
AI总结：
本文探讨了在大视觉语言模型中思考与回顾的平衡，提出不确定性引导的回顾策略，通过大规模对比实验提升MMMU性能，并在多个基准上取得新突破。
摘要：
测试时思考（即生成显式的中间推理链）已被证明能提升大语言模型的性能，并在大视觉语言模型（LVLMs）中表现出显著优势。然而，尽管这些结果令人鼓舞，目前仍缺乏对思考如何实际影响视觉推理的系统分析。本文通过大规模、受控的比较分析，评估了十种LVLM变体在MMMU-val上的表现，采用慷慨的token预算和多轮解码。研究发现，更多的思考并不总是更好；长链常导致长错误轨迹，忽视图像内容并低于标准指令模式下的表现。深入分析表明，某些明确指向图像的短回顾短语在成功轨迹中富集，并与更好的视觉基础相关。基于此见解，本文提出不确定性引导的回顾策略，一种无需训练的解码策略，结合不确定性信号与自适应回顾提示和广度搜索。该方法提升了整体MMMU性能，在标准思考薄弱的类别中实现了最大的提升，并超越了几个强大的解码基线，在固定模型家族和token预算下设定了新的基准。此外，本文还显示该解码策略具有泛化能力，在五个额外基准上持续提升，包括两个广泛的多模态套件和专注于数学的视觉推理数据集。
英文摘要：
Test-time thinking (that is, generating explicit intermediate reasoning chains) is known to boost performance in large language models and has recently shown strong gains for large vision language models (LVLMs). However, despite these promising results, there is still no systematic analysis of how thinking actually affects visual reasoning. We provide the first such analysis with a large scale, controlled comparison of thinking for LVLMs, evaluating ten variants from the InternVL3.5 and Qwen3-VL families on MMMU-val under generous token budgets and multi pass decoding. We show that more thinking is not always better; long chains often yield long wrong trajectories that ignore the image and underperform the same models run in standard instruct mode. A deeper analysis reveals that certain short lookback phrases, which explicitly refer back to the image, are strongly enriched in successful trajectories and correlate with better visual grounding. Building on this insight, we propose uncertainty guided lookback, a training free decoding strategy that combines an uncertainty signal with adaptive lookback prompts and breadth search. Our method improves overall MMMU performance, delivers the largest gains in categories where standard thinking is weak, and outperforms several strong decoding baselines, setting a new state of the art under fixed model families and token budgets. We further show that this decoding strategy generalizes, yielding consistent improvements on five additional benchmarks, including two broad multimodal suites and math focused visual reasoning datasets.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
