CADS: Unleashing the Diversity of Diffusion Models through Condition-Annealed Sampling

TL;DR

This paper introduces CADS, a novel sampling method that enhances diversity in conditional diffusion models by annealing the conditioning signal, leading to improved image generation quality and diversity, especially at high guidance scales.

Contribution

CADS is a new sampling strategy that improves diversity in diffusion models by annealing conditioning signals, compatible with any pretrained model and sampling algorithm.

Findings

CADS boosts diversity in various conditional generation tasks.

Achieves state-of-the-art FID scores of 1.70 and 2.31 on ImageNet at 256x256 and 512x512.

Enhances output diversity without significant loss of sample quality.

Abstract

While conditional diffusion models are known to have good coverage of the data distribution, they still face limitations in output diversity, particularly when sampled with a high classifier-free guidance scale for optimal image quality or when trained on small datasets. We attribute this problem to the role of the conditioning signal in inference and offer an improved sampling strategy for diffusion models that can increase generation diversity, especially at high guidance scales, with minimal loss of sample quality. Our sampling strategy anneals the conditioning signal by adding scheduled, monotonically decreasing Gaussian noise to the conditioning vector during inference to balance diversity and condition alignment. Our Condition-Annealed Diffusion Sampler (CADS) can be used with any pretrained model and sampling algorithm, and we show that it boosts the diversity of diffusion models in various conditional generation tasks. Further, using an existing pretrained diffusion model, CADS achieves a new state-of-the-art FID of 1.70 and 2.31 for class-conditional ImageNet generation at 256$\times$256 and 512$\times$512 respectively.