DALL·E 2, DALL·E 3, Midjourney, and Imagen

DALL·E 2, DALL·E 3, Midjourney, and Imagen are text-to-image models used in the paper for training and evaluation.

BibTex: