重新审视以视觉为中心的推理泛化中长思维链的必要性
Revisiting the Necessity of Lengthy Chain-of-Thought in Vision-centric Reasoning Generalization
对做视觉推理 SFT 数据构造的人有直接指导：不必堆长 CoT 或图像操作式 CoT，只保留关键视觉定位步骤的极简 CoT 反而在不同难度（网格尺寸）上泛化最好，可作为构建可泛化视觉推理训练数据的默认策略。
原文依据 CoT retaining only the minimal grounding results generalizes best across different maze sizes
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论本文研究不同的思维链（CoT）设计如何影响视觉语言模型（VLM）获得可泛化视觉推理能力。尽管 CoT 数据——尤其是长 CoT 或视觉 CoT（如“用图像思考”）——已被广泛用于监督中间推理过程，但特定 CoT 设计为何有效、哪些设计真正支持可泛化推理，仍不清楚。为系统评估这一问题，我们聚焦于一个受控的迷宫求解任务：其推理规则完全以视觉形式呈现，难度可通过网格尺寸调节，且所有中间步骤均可自动生成。我们在标准的“先监督微调后强化学习”流程下使用 Qwen2.5-VL-7B，比较三种代表性 CoT 格式：语言 CoT、视觉定位 CoT（带空间坐标轨迹）与视觉 CoT（带图像操作）。实验表明：视觉 CoT 与更长的 CoT 主要加速收敛，但并不提升最终性能上限；仅包含必要视觉定位步骤的简洁 CoT 优于更长的推理轨迹；尤为突出的是，仅保留最小视觉定位信息的 CoT 在不同迷宫尺寸上泛化最好。我们进一步在其他以视觉为中心的任务上验证了这些结论。这些发现凸显出一种“短即是长”效应，并为构建更具泛化性的视觉推理监督微调数据集提供了实用指导。我们的代码与数据将在 https://github.com/RUCAIBox/Revisiting-Visual-CoT 公开。
英文原文摘要 1459 字符
We study how different Chain-of-Thought (CoT) designs affect the acquisition of the generalizable visual reasoning ability in vision-language models (VLMs). While CoT data, especially long or visual CoT such as "think with image", has been widely used to supervise intermediate reasoning, it remains unclear why specific CoT designs help and which ones truly support generalizable reasoning. To systematically evaluate this, we focus on a controlled maze-solving task where reasoning rules are fully visual, difficulty can be tuned by grid size, and all the intermediate steps can be automatically generated. Using Qwen2.5-VL-7B under a standard SFT-then-RL pipeline, we compare three representative CoT formats: Language CoT, Grounding CoT (with spatial coordinate trajectories), and Visual CoT (with image manipulations). Our experiments reveal that visual and longer CoT mainly accelerate convergence but do not lift the final performance ceiling; concise CoT containing only essential grounding steps outperforms longer traces; and, strikingly, CoT retaining only the minimal grounding results generalizes best across different maze sizes. We further validate these insights on other vision-centric tasks. These findings highlight a "short is long" effect and provide practical guidance for constructing more generalizable SFT datasets for visual reasoning. Our code and data will be publicly released at https://github.com/RUCAIBox/Revisiting-Visual-CoT.