The dataset consists of public social media url pairs and the corresponding entailment label for an external conference (ACL 2021). Each url contains a post with both linguistic (text) and visual (image) content. Entailment labels are human annotated through Google Crowdsource.