Socratic-Geo：通过多智能体交互的合成数据生成与跨模态几何推理
Socratic-Geo: Synthetic Data Generation and Cross-Modal Geometric Reasoning via Multi-Agent Interaction
对做视觉语言模型几何推理的人，这篇提供了用多智能体闭环把「图像-文本对」合成与模型学习需求耦合起来的范式：Teacher 用参数化 Python 脚本加可解性/视觉有效性反思保证数据纯度，Solver 的失败路径反过来指导定向增广，Generator 把「图像-代码-指令」三元组蒸馏成图像生成能力，可直接迁移到其他数据稀缺的推理任务。
原文依据 The Solver agent optimizes reasoning through preference learning, with failure paths guiding Teacher's targeted augmentation.
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论多模态大语言模型（MLLMs）已显著推进视觉-语言理解。然而，即便最优模型仍在几何推理上表现吃力，这揭示出一个关键瓶颈：高质量图文对的极度稀缺。人工标注成本高得难以承受，而自动化方法又无法同时保证保真度与训练有效性。现有方法要么被动地适配已有图像，要么采用低效的随机探索加事后过滤，使生成与学习需求相互脱节。我们提出 Socratic-Geo，一个完全自主的框架，通过多智能体交互将数据合成与模型学习动态耦合。Teacher 智能体生成参数化的 Python 脚本并附带反思式反馈（Reflect 用于可解性，RePI 用于视觉有效性），以确保图文对的纯度。Solver 智能体通过偏好学习优化推理，其失败路径用于指导 Teacher 的定向增广。Generator 则独立地在累积的「图像-代码-指令」三元组上学习图像生成能力，将程序化绘图智能蒸馏进视觉生成。仅从 108 个种子问题出发，Socratic-Solver 在使用四分之一基线数据的情况下，在六个几何基准上取得 49.11，超过强基线 2.43 分。Socratic-Generator 在 GenExam 上取得 42.4%，为开源模型树立了新的最优，超过 Seedream-4.0（39.8%），并接近 Gemini-2.5-Flash-Image（43.1%）。
英文原文摘要 1545 字符
Multimodal Large Language Models (MLLMs) have significantly advanced vision-language understanding. However, even state-of-the-art models struggle with geometric reasoning, revealing a critical bottleneck: the extreme scarcity of high-quality image-text pairs. Human annotation is prohibitively expensive, while automated methods fail to ensure both fidelity and training effectiveness. Existing approaches either passively adapt to available images or employ inefficient random exploration with post-hoc filtering, decoupling generation from learning needs. We propose Socratic-Geo, a fully autonomous framework that dynamically couples data synthesis with model learning through multi-agent interaction. The Teacher agent generates parameterized Python scripts with reflective feedback (Reflect for solvability, RePI for visual validity), ensuring image-text pair purity. The Solver agent optimizes reasoning through preference learning, with failure paths guiding Teacher's targeted augmentation. Independently, the Generator learns image generation capabilities on accumulated "image-code-instruction" triplets, distilling programmatic drawing intelligence into visual generation. Starting from only 108 seed problems, Socratic-Solver achieves 49.11 on six geometric benchmarks using one-quarter of baseline data, surpassing strong baselines by 2.43 points. Socratic-Generator achieves 42.4% on GenExam, establishing new state-of-the-art for open-source models, surpassing Seedream-4.0 (39.8%) and approaching Gemini-2.5-Flash-Image (43.1%).