loganbolton/sketchvlm-connect-dots

该数据集是SketchVLM框架的组成部分，专门用于评估视觉语言模型(VLMs)预测球通过障碍物轨迹的能力(基于PHYRE)。SketchVML框架使VLMs能够在输入图像上生成非破坏性、可编辑的SVG覆盖层，以直观解释其推理过程。

This dataset is part of the SketchVLM framework, which enables Vision-Language Models (VLMs) to produce non-destructive, editable SVG overlays on input images to visually explain their reasoning. The Physics Ball Drop benchmark specifically evaluates a models ability to predict the trajectory of a ball through obstacles (based on PHYRE).