Elucidating the Design Space of Diffusion-Based Generative Models

06/01/2022
by   Tero Karras, et al.
0

We argue that the theory and practice of diffusion-based generative models are currently unnecessarily convoluted and seek to remedy the situation by presenting a design space that clearly separates the concrete design choices. This lets us identify several changes to both the sampling and training processes, as well as preconditioning of the score networks. Together, our improvements yield new state-of-the-art FID of 1.79 for CIFAR-10 in a class-conditional setting and 1.97 in an unconditional setting, with much faster sampling (35 network evaluations per image) than prior designs. To further demonstrate their modular nature, we show that our design changes dramatically improve both the efficiency and quality obtainable with pre-trained score networks from previous work, including improving the FID of an existing ImageNet-64 model from 2.07 to near-SOTA 1.55.

READ FULL TEXT

page 5

page 13

page 14

page 15

page 16

page 17

page 18

page 40

research
02/09/2023

Geometry of Score Based Generative Models

In this work, we look at Score-based generative models (also called diff...
research
06/10/2021

Score-based Generative Modeling in Latent Space

Score-based generative models (SGMs) have recently demonstrated impressi...
research
09/26/2022

Quasi-Conservative Score-based Generative Models

Existing Score-based Generative Models (SGMs) can be categorized into co...
research
07/09/2022

Improving Diffusion Model Efficiency Through Patching

Diffusion models are a powerful class of generative models that iterativ...
research
12/01/2022

Why Are Conditional Generative Models Better Than Unconditional Ones?

Extensive empirical evidence demonstrates that conditional generative mo...
research
03/03/2023

Generative Diffusions in Augmented Spaces: A Complete Recipe

Score-based Generative Models (SGMs) have achieved state-of-the-art synt...
research
06/15/2023

OMS-DPM: Optimizing the Model Schedule for Diffusion Probabilistic Models

Diffusion probabilistic models (DPMs) are a new class of generative mode...

Please sign up or login with your details

Forgot password? Click here to reset