Pre-trained Language Models Do Not Help Auto-regressive Text-to-Image Generation.

Yuhui Zhang, Brandon McKinzie, Zhe Gan, Vaishaal Shankar, Alexander Toshev

Browse the full EMNLP paper archive.