AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Using Diffusion to Generate High Res Images
Open AI has been using a technique called diffusion to generate images. The process is much more efficient, both in terms of the amount of training compute required to train the models as well as in the number of steps required to generate the images. There are many interesting use cases for dolly including people typing their favorite painter and then giving them some obscure modern topic.