Hybrid SD: Edge-Cloud Collaborative Inference for Stable Diffusion Models
This work addresses the computational and cost barriers for deploying large image synthesis models on edge devices, offering a practical solution for applications requiring efficient AI inference, though it is incremental in combining existing techniques like pruning and collaborative inference.
The paper tackles the problem of deploying Stable Diffusion Models (SDMs) on edge devices by introducing Hybrid SD, a training-free edge-cloud collaborative inference framework that distributes early diffusion steps to the cloud for semantic planning and later steps to edge devices for visual refinement, achieving state-of-the-art parameter efficiency (225.8M) and reducing cloud cost by 66%.
Stable Diffusion Models (SDMs) have shown remarkable proficiency in image synthesis. However, their broad application is impeded by their large model sizes and intensive computational requirements, which typically require expensive cloud servers for deployment. On the flip side, while there are many compact models tailored for edge devices that can reduce these demands, they often compromise on semantic integrity and visual quality when compared to full-sized SDMs. To bridge this gap, we introduce Hybrid SD, an innovative, training-free SDMs inference framework designed for edge-cloud collaborative inference. Hybrid SD distributes the early steps of the diffusion process to the large models deployed on cloud servers, enhancing semantic planning. Furthermore, small efficient models deployed on edge devices can be integrated for refining visual details in the later stages. Acknowledging the diversity of edge devices with differing computational and storage capacities, we employ structural pruning to the SDMs U-Net and train a lightweight VAE. Empirical evaluations demonstrate that our compressed models achieve state-of-the-art parameter efficiency (225.8M) on edge devices with competitive image quality. Additionally, Hybrid SD reduces the cloud cost by 66% with edge-cloud collaborative inference.