Fun fact: common text-to-image Diffusion models actually use Transformers

→ View original post on X — @rasbt