loganbolton/sketchvlm-maze-navigation
收藏Hugging Face2026-04-29 更新2026-05-03 收录
下载链接：
https://hf-mirror.com/datasets/loganbolton/sketchvlm-maze-navigation
下载链接
链接失效反馈官方服务：
资源简介：
SketchVLM: Maze Navigation数据集是与论文[SketchVLM: Vision Language Models Can Annotate Images to Explain Thoughts and Guide Users]相关的数据集。该数据集用于评估模型在迷宫中从起点到终点追踪路径的能力。数据集包含多个特征，如迷宫图像、带路径的迷宫图像、迷宫ID、路径长度、最短路径方向、最短路径坐标等。数据集的大小为2212549.0字节，包含200个训练样本。
The SketchVLM: Maze Navigation dataset is associated with the paper: [SketchVLM: Vision Language Models Can Annotate Images to Explain Thoughts and Guide Users]. This dataset is one of the benchmarks introduced to evaluate a models ability to trace a path from start to end through a maze. The dataset includes features such as maze images, maze images with paths, maze IDs, path lengths, shortest path directions, shortest path coordinates, etc. The dataset size is 2212549.0 bytes, containing 200 training examples.
提供机构：
loganbolton
