SHOW-O: ONE SINGLE TRANSFORMER TO UNIFY MULTIMODAL UNDERSTANDING AND GENERATION

Summary from the TLDR AI newsletter

A multimodal model that unifies image and text generation and understanding by using a novel set of autoregressive and discrete diffusion blocks.