Modulated Diffusion: Accelerating Generative Modeling with Modulated Quantization

TL;DR

This paper introduces Modulated Diffusion (MoDiff), a novel framework that accelerates diffusion models using modulated quantization and error compensation, reducing computational costs while maintaining high-quality generative performance.

Contribution

MoDiff provides a new, principled approach to accelerate diffusion models through modulated quantization and error correction, surpassing limitations of existing methods.

Findings

Reduces activation quantization from 8 bits to 3 bits without performance loss

Demonstrates significant speedup in diffusion sampling processes

Applicable to various diffusion models with theoretical support

Abstract

Diffusion models have emerged as powerful generative models, but their high computation cost in iterative sampling remains a significant bottleneck. In this work, we present an in-depth and insightful study of state-of-the-art acceleration techniques for diffusion models, including caching and quantization, revealing their limitations in computation error and generation quality. To break these limits, this work introduces Modulated Diffusion (MoDiff), an innovative, rigorous, and principled framework that accelerates generative modeling through modulated quantization and error compensation. MoDiff not only inherents the advantages of existing caching and quantization methods but also serves as a general framework to accelerate all diffusion models. The advantages of MoDiff are supported by solid theoretical insight and analysis. In addition, extensive experiments on CIFAR-10 and LSUN demonstrate that MoDiff significant reduces activation quantization from 8 bits to 3 bits without performance degradation in post-training quantization (PTQ). Our code implementation is available at https://github.com/WeizhiGao/MoDiff.