1 paper
Shaghayegh Kolli, Sina Emami, Moreno D'Incà +4
Text-to-image models learn associations between concepts - in the case of this paper, people's professions, which we refer to as roles - and visual attributes. These associations c…