pomelyu / paper-reading-notes

0 stars 0 forks source link

2021 [NIPS] (CFG) Classifier-Free Diffusion Guidance #5

Open pomelyu opened 1 year ago

pomelyu commented 1 year ago

Introduction

The CFG factor used in stable-diffusion. Unlike truncation trick in GAN and temperature in FLOW model, diffusion model doesn't have the way to control the trade-off between the image quality and the variety. As a result, Diffusion models beat GANs on image synthesis[^1] introduced Class Guidance, which use the gradient from the pre-trained classifier in the inference time to obtain this trade-off. However, Class Guidance has some drawbacks, and this paper proposed an alternative way that does not need the classifier to calculate the class guidance.

Method

  1. In detail, here is the class guidance

    image
  2. after we apply the weight $w+1$ to unconditional model, we get conditional model with weight $w$

    image
  3. In inference time

    image

Highlight

Limitation

Comments

[^1]: Diffusion models beat GANs on image synthesis