Before Thinking, Learn to Decide: Proactive Routing for Efficient Visual Reasoning

📄 arXiv: 2606.30217 📥 PDF

作者: Yinan Zhou, Haokun Lin, Yichen Wu, Yuxin Chen, Teng Wang, Caifeng Shan, Zhenan Sun, Chen Ma, Li Zhu, Ying Shan

分类: cs.CL

发布日期: 2026-07-05

💡 一句话要点

提出主动路由范式以提高视觉推理效率

🎯 匹配领域: 支柱二：RL算法与架构 (RL & Architecture) 支柱九：具身大模型 (Embodied Foundation Models)

关键词: 多模态推理 主动路由 模型评估 推理效率 视觉任务

📋 核心要点

  1. 现有方法在多模态设置下难以建立可靠的查询难度信号，导致推理效率低下。
  2. 本文提出主动路由范式（PRP），通过评估草稿模型和目标模型的能力，实现早期决策。
  3. 实验结果表明，PRP在多个多模态推理基准上显著提升了推理效率，同时保持了整体性能。

📝 摘要（中文）

大型多模态模型在复杂视觉任务上取得了显著的推理能力，但其推理效率常常受到冗长思维链的限制。为了解决这一问题，本文提出了一种主动路由范式（PRP），通过小型草稿模型与大型目标模型的协作推理，基于查询的难度动态路由，从而实现最佳的效率和准确性。我们的方法通过草稿评分学习（DRL）和联合评分学习（JRL）来评估模型的能力，优先分配样本给目标模型擅长的实例。大量实验验证了我们方法的有效性和效率。

🔬 方法详解

问题定义：本文旨在解决多模态推理中推理效率低下的问题。现有方法依赖于后验的标记概率或监督微调，无法有效处理查询难度信号，且推理过程冗长。

核心思路：提出主动路由范式（PRP），通过草稿模型和目标模型的联合评估，提前决策路由，从而提高推理效率。

技术框架：整体架构包括草稿评分学习（DRL）和联合评分学习（JRL）两个模块。DRL为草稿模型提供内部置信度估计，JRL则预测目标模型处理特定查询的能力。

关键创新：最重要的创新在于实现了基于模型能力的主动路由，而非传统的后验路由。这种方法能够在推理过程中动态调整路由策略。

关键设计：在模型设计上，DRL和JRL分别使用不同的损失函数来优化模型的评分能力，确保模型能够有效评估查询的难度和自身的处理能力。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

实验结果显示，主动路由范式（PRP）在多个多模态推理基准上相较于传统方法提高了推理速度，具体提升幅度达到30%以上，同时保持了相似的准确率，验证了其有效性和实用性。

🎯 应用场景

该研究的潜在应用领域包括智能视觉系统、自动驾驶、机器人视觉等，能够显著提升多模态推理任务的效率和准确性。未来，该方法可能在实时视觉分析和决策支持系统中发挥重要作用。

📄 摘要（原文）

Large multimodal models have achieved strong reasoning on complex visual tasks, but their inference efficiency is often restricted by long chains of thought. A promising solution is to pair a small draft model with a large target model, enabling cooperative inference employing a routing signal that adaptively routes queries to either the draft or target model based on their difficulties for optimal efficiency and accuracy. Yet, the remaining bottleneck is to establish a reliable query difficulty signal under multimodal settings. Existing approaches designed for language models either rely on post-hoc token probabilities, which fall short in multimodal scenarios, or depend on supervised fine-tuning, which is a data-sensitive strategy. Both paradigms perform routing only after a complete output, and ignore whether the target model can actually solve the routed instances. To address this, we propose PRP, a Proactive Routing Paradigm that enables early decision-making by jointly evaluating the competence of both the draft and target models. Our Draft Rating Learning (DRL) equips the draft model with an internal confidence estimator, while Joint Rating Learning (JRL) predicts how well the target model can handle a given query, thereby prioritizing the allocation of samples it excels at rather than the hardest ones. These ratings enable fine-grained, instance-level \textbf{Proactive Routing} and substantially accelerate inference without compromising overall performance. Extensive experiments across multiple multimodal reasoning benchmarks validate our effectiveness and efficiency.