Attention-Driven Training-Free Efficiency Enhancement of Diffusion Models

TL;DR

This paper introduces a run-time, training-free token pruning method for diffusion models that significantly reduces computational costs while preserving high image quality, using attention maps and novel algorithms.

Contribution

It proposes AT-EDM, a novel framework that leverages attention maps for efficient, training-free token pruning in diffusion models, including new algorithms G-WPR and DSAP for improved performance.

Findings

38.8% FLOPs reduction achieved

Up to 1.53x speed-up over Stable Diffusion XL

Maintains similar FID and CLIP scores as full models

Abstract

Diffusion Models (DMs) have exhibited superior performance in generating high-quality and diverse images. However, this exceptional performance comes at the cost of expensive architectural design, particularly due to the attention module heavily used in leading models. Existing works mainly adopt a retraining process to enhance DM efficiency. This is computationally expensive and not very scalable. To this end, we introduce the Attention-driven Training-free Efficient Diffusion Model (AT-EDM) framework that leverages attention maps to perform run-time pruning of redundant tokens, without the need for any retraining. Specifically, for single-denoising-step pruning, we develop a novel ranking algorithm, Generalized Weighted Page Rank (G-WPR), to identify redundant tokens, and a similarity-based recovery method to restore tokens for the convolution operation. In addition, we propose a Denoising-Steps-Aware Pruning (DSAP) approach to adjust the pruning budget across different denoising timesteps for better generation quality. Extensive evaluations show that AT-EDM performs favorably against prior art in terms of efficiency (e.g., 38.8% FLOPs saving and up to 1.53x speed-up over Stable Diffusion XL) while maintaining nearly the same FID and CLIP scores as the full model. Project webpage: https://atedm.github.io.