Contrastive Visual Semantic Pretraining Magnifies the Semantics of Natural Language Representations

Robert Wolfe, Aylin Caliskan