从生成视角探索空间智能
Exploring Spatial Intelligence from a Generative Perspective
提供了一个可复用的评测协议与合成数据管线：用 3D 先验引导生成并自动标注空间编辑样本，可直接用于评测或微调统一多模态模型的 3D 空间约束遵循能力，并验证生成式训练反哺空间理解。
原文依据 We introduce GSI-Bench, the first benchmark designed to quantify GSI through spatially grounded image editing.
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论空间智能对多模态大语言模型至关重要，然而当前基准大多只从理解的角度对其进行评估。我们提出疑问：现代生成式或统一多模态模型是否也具备生成式空间智能（GSI），即在图像生成过程中遵循并操控三维空间约束的能力，以及这种能力能否被度量或提升。我们提出 GSI-Bench，这是首个旨在通过空间接地的图像编辑来量化 GSI 的基准。它由两个互补部分组成：GSI-Real，一个通过三维先验引导的生成与筛选流程构建的高质量真实世界数据集；以及 GSI-Syn，一个具有可控空间操作与全自动标注的大规模合成基准。结合统一的评测协议，GSI-Bench 能够以可扩展、与模型无关的方式评估空间遵循性与编辑保真度。实验表明，在 GSI-Syn 上微调统一多模态模型可在合成与真实任务上均取得显著提升，并且引人注目的是，还能改善下游的空间理解能力。这提供了首个明确证据，表明生成式训练能够切实增强空间推理，为推进多模态模型的空间智能开辟了新路径。
英文原文摘要 1313 字符
Spatial intelligence is essential for multimodal large language models, yet current benchmarks largely assess it only from an understanding perspective. We ask whether modern generative or unified multimodal models also possess generative spatial intelligence (GSI)--the ability to respect and manipulate 3D spatial constraints during image generation--and whether such capability can be measured or improved. We introduce GSI-Bench, the first benchmark designed to quantify GSI through spatially grounded image editing. It consists of two complementary components: GSI-Real, a high-quality real-world dataset built via a 3D-prior-guided generation and filtering pipeline, and GSI-Syn, a large-scale synthetic benchmark with controllable spatial operations and fully automated labeling.Together with a unified evaluation protocol, GSI-Bench enables scalable, model-agnostic assessment of spatial compliance and editing fidelity. Experiments show that fine-tuning unified multimodal models on GSI-Syn yields substantial gains on both synthetic and real tasks and, strikingly, also improves downstream spatial understanding. This provides the first clear evidence that generative training can tangibly strengthen spatial reasoning--establishing a new pathway for advancing spatial intelligence in multimodal models.