VisionLeaf：面向高效准确 think-with-image 的熵引导叶优先推理
VisionLeaf: Entropy-Guided Leaf-First Reasoning for Efficient and Accurate Think-with-Image
把 GRPO 的树式 rollout 从根节点扩展改为从叶节点按熵选择扩展，可直接迁移到任何多轮视觉推理 VLM 的强化学习训练中，在不改模型与数据的前提下同时降低推理轮数并提升准确率。
原文依据 our method grows the reasoning tree from the leaf nodes and selects the most valuable nodes based on entropy for thorough rollout exploration
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论“think-with-image”范式近期在复杂视觉推理任务中受到关注。然而，现有方法常因固定数量的冗余推理步骤而面临推理效率低下的问题，同时训练也不稳定。这一挑战主要源于直接使用标准的强化学习策略，而这些策略并未针对 think-with-image 的多轮对话场景做出改进。为解决该问题，我们提出 VisionLeaf，一个熵引导的、基于树的推理框架。与所有节点均从根节点扩展、且每个叶节点只有单一分支的传统 GRPO 不同，我们的方法从叶节点生长推理树，并基于熵选择最有价值的节点以进行充分的 rollout 探索。这种叶优先的扩展方式天然契合多步图像分析的分层特性。在不修改任何模型或训练数据的情况下，VisionLeaf 在 VSTAR、HRBench 等基准上取得 4.2% 的性能提升，同时将推理轮数减少近一半——在准确率与速度上均展现出显著收益。我们的代码将全部开源。
英文原文摘要 1193 字符
The "think-with-image" paradigm has recently gained traction for complex visual reasoning tasks. However, existing approaches often struggle with inference inefficiency due to a fixed number of redundant reasoning steps, as well as training instability. This challenge primarily arises from the direct use of standard reinforcement learning policies, which do not incorporate improvements for the think-with-image multi-turn conversational scenario. To address this challenge, we propose VisionLeaf, an entropy-guided, tree-based reasoning framework. Unlike conventional GRPO, where all nodes expand from the root and each leaf has only a single branch, our method grows the reasoning tree from the leaf nodes and selects the most valuable nodes based on entropy for thorough rollout exploration. This leaf-first expansion naturally aligns with the hierarchical nature of multi-step image analysis. Without modifying any model or training data, our VisionLeaf achieves a 4.2% performance improvement on benchmarks such as VSTAR and HRBench, while reducing the number of inference rounds by nearly half--demonstrating significant gains in both accuracy and speed. All our code will be released.