1 paper
Bo Yang, Fangxiang Feng, Xiaojie Wang
A good Text-to-Image model should not only generate high quality images, but also ensure the consistency between the text and the generated image. Previous models failed to simulta…