On Fast Sampling of Diffusion Probabilistic Models

TL;DR

FastDPM introduces a unified framework for rapid sampling in diffusion probabilistic models, enhancing sample quality and offering adaptable algorithms for various data types and conditional settings.

Contribution

The paper presents FastDPM, a comprehensive framework that generalizes existing methods and introduces new algorithms for faster sampling in diffusion models.

Findings

Performance varies with data domain and conditional information.

Trade-off exists between sampling speed and sample quality.

Guidelines provided for method selection in practice.

Abstract

In this work, we propose FastDPM, a unified framework for fast sampling in diffusion probabilistic models. FastDPM generalizes previous methods and gives rise to new algorithms with improved sample quality. We systematically investigate the fast sampling methods under this framework across different domains, on different datasets, and with different amount of conditional information provided for generation. We find the performance of a particular method depends on data domains (e.g., image or audio), the trade-off between sampling speed and sample quality, and the amount of conditional information. We further provide insights and recipes on the choice of methods for practitioners.