SPARTQA
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/hlr/spartqa_generation
下载链接
链接失效反馈官方服务：
资源简介：
该数据集旨在评估基础空间推理能力。在此数据集中，模型在处理基础空间推理任务时表现尚可，但当空间推理的复杂性增加时，模型的表现则显得较为吃力。所涉及的任务是空间推理。
This dataset is designed to evaluate basic spatial reasoning capabilities. In this dataset, models perform moderately well when handling basic spatial reasoning tasks, but struggle considerably when the complexity of such spatial reasoning tasks increases. The tasks involved are spatial reasoning.
