DATASET

The Hexagons dataset comprises 4177 naturally-occurring visually grounded instructions rich with diverse types and levels of abstractions.
The dataset as well as the evaluation of its quality is described in the paper.

The following table summarizes some of its properties.

The dataset is split into train, dev, and test according to the following table and may be downloaded here.