Geometry3k Geometry Problem Dataset

The Geometry3k dataset is a geometric problem dataset released by the University of California, Los Angeles and Zhejiang University. It is mainly used for visual question answering tasks. The relevant paper results are:Inter-GPS: Interpretable Geometry Problem Solving with Formal Language and Symbolic ReasoningThe dataset has a total of 3002 samples, including image and text data, which are divided into training set, validation set and test set in a ratio of 7:1:2, of which the training set has about 2100 data.

This dataset describes a rich variety of geometric problems, such as solving angles, side lengths, areas, perimeters, etc., with a total of 6293 text items.

Diagrams are used to assist in presenting graphical information in geometry problems, such as various geometric shapes (triangles, circles, quadrilaterals, etc.) and their relationships, with a total of 27,213 text items.