A sizable data set made up of pairs of photos and their related text descriptions is used to train DALL-E. The link between visual information and written representation is taught to the model ...