Zero-Shot Text-to-Image Generation

               Aditya Ramesh 1 Mikhail Pavlov 1 Gabriel Goh 1 Scott Gray 1
               Chelsea Voss 1 Alec Radford 1 Mark Chen 1 Ilya Sutskever 1

             Abstract
    Text-to-image generation has traditionally fo-
    cused on finding better modeling assumptions for
    training on a fixed dataset. These assumptions
    might involve complex architectures, auxiliary
    losses, or side information such as object part la-
    bels or segmentation masks ...