e-SNLI-VE is a large-scale visual-linguistic dataset with natural language explanations, containing more than 430,000 instances, all of which are based on image content. The dataset is built by merging the explanations from the e-SNLI dataset and the image-sentence pairs from the SNLI-VE dataset.

Citation

@InProceedings{Kayser_2021_ICCV,
author    = {Kayser, Maxime and Camburu, Oana-Maria and Salewski, Leonard and Emde, Cornelius and Do, Virginie and Akata, Zeynep and Lukasiewicz, Thomas},
title     = {E-ViL: A Dataset and Benchmark for Natural Language Explanations in Vision-Language Tasks},
booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
month     = {October},
year      = {2021},
pages     = {1244-1254}
}