What is the architecture of text-to-image generation?

[weixiang-zhang]

Dear authors,

Great work. I am wondering what is the t2i architecture of this work? MMDit or pixel-art style? is there any detailed descriptions?

Thanks!