Squeezing Large-Scale Diffusion Models for Mobile

TL;DR

This paper addresses the challenge of deploying large diffusion models like Stable Diffusion on mobile devices by proposing solutions that enable efficient on-device inference, achieving under 7 seconds latency for image generation.

Contribution

The paper introduces methods for compressing and optimizing large diffusion models for mobile deployment using TensorFlow Lite, facilitating practical on-device image synthesis.

Findings

Mobile Stable Diffusion achieves inference in under 7 seconds on Android devices.

The approach enables deployment of models with over one billion parameters on resource-constrained devices.

The solutions support both iOS and Android platforms.

Abstract

The emergence of diffusion models has greatly broadened the scope of high-fidelity image synthesis, resulting in notable advancements in both practical implementation and academic research. With the active adoption of the model in various real-world applications, the need for on-device deployment has grown considerably. However, deploying large diffusion models such as Stable Diffusion with more than one billion parameters to mobile devices poses distinctive challenges due to the limited computational and memory resources, which may vary according to the device. In this paper, we present the challenges and solutions for deploying Stable Diffusion on mobile devices with TensorFlow Lite framework, which supports both iOS and Android devices. The resulting Mobile Stable Diffusion achieves the inference latency of smaller than 7 seconds for a 512x512 image generation on Android devices with mobile GPUs.