Improving Efficiency of Diffusion Models via Multi-Stage Framework and Tailored Multi-Decoder Architectures

TL;DR

This paper introduces a multi-stage framework with tailored multi-decoder architectures to improve the training and sampling efficiency of diffusion models, addressing their computational challenges while maintaining high performance.

Contribution

The paper proposes a novel multi-stage approach with custom multi-decoder U-net architectures and a timestep clustering algorithm to enhance diffusion models' efficiency.

Findings

Significant improvements in training speed across three state-of-the-art diffusion models.

Enhanced sampling efficiency without sacrificing generative quality.

Effective resource distribution and reduced inter-stage interference.

Abstract

Diffusion models, emerging as powerful deep generative tools, excel in various applications. They operate through a two-steps process: introducing noise into training samples and then employing a model to convert random noise into new samples (e.g., images). However, their remarkable generative performance is hindered by slow training and sampling. This is due to the necessity of tracking extensive forward and reverse diffusion trajectories, and employing a large model with numerous parameters across multiple timesteps (i.e., noise levels). To tackle these challenges, we present a multi-stage framework inspired by our empirical findings. These observations indicate the advantages of employing distinct parameters tailored to each timestep while retaining universal parameters shared across all time steps. Our approach involves segmenting the time interval into multiple stages where we employ custom multi-decoder U-net architecture that blends time-dependent models with a universally shared encoder. Our framework enables the efficient distribution of computational resources and mitigates inter-stage interference, which substantially improves training efficiency. Extensive numerical experiments affirm the effectiveness of our framework, showcasing significant training and sampling efficiency enhancements on three state-of-the-art diffusion models, including large-scale latent diffusion models. Furthermore, our ablation studies illustrate the impact of two important components in our framework: (i) a novel timestep clustering algorithm for stage division, and (ii) an innovative multi-decoder U-net architecture, seamlessly integrating universal and customized hyperparameters.