1 dataset found

Tags: CIRR

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.
You can also access this registry using the API (see API Docs).