Fetching the paper…

Winoground: Probing Vision and Language Models for Visio-Linguistic Compositionality · Around