📃➡️🖼 Edge#227: Autoregressive Text-to-Image Models
+Google's Parti; +MS COCO
In this issue:
we explain autoregressive text-to-image models;
we discuss Google’s Parti, an impressive autoregressive text-to-image model;
we explore MS COCO, one of the most common datasets in text-to-image models.
Enjoy the learning!
💡 ML Concept of the Day: Autoregressive Text-to-Image Models
The first editions of our text-to-image series have focused on exploring diffusion models, which have been at the forefront of some of the most impressive developments in this area. Diffusion models work by introducing Gaussian noise into an input image and then training a model to reconstruct the original image from the noise. An interesting alternative to diffusion models is autoregressive methods used for text-to-image synthesis.
The core idea of autoregressive methods is