1 dataset found

Groups: Real-Life Images

Filter Results

  • CIRR

    CIRR is a general image dataset that comprises 36,554 triplets derived from 21,552 images from the popular natural language inference dataset NLVR2.