SnapGen: Taming High-Resolution Text-to-Image Models for Mobile Devices with Efficient Architectures and Training

TL;DR

SnapGen introduces a highly efficient, small, and fast text-to-image diffusion model capable of generating high-resolution images on mobile devices, surpassing larger models in quality and speed.

Contribution

The paper presents novel architecture design, knowledge distillation, and adversarial guidance techniques to enable high-quality, high-resolution image generation on mobile with a tiny model.

Findings

Generates 1024x1024 images on mobile in 1.4 seconds.

Achieves an FID of 2.06 on ImageNet-1K with only 372M parameters.

Outperforms larger models like SDXL and IF-XL in size and quality.

Abstract

Existing text-to-image (T2I) diffusion models face several limitations, including large model sizes, slow runtime, and low-quality generation on mobile devices. This paper aims to address all of these challenges by developing an extremely small and fast T2I model that generates high-resolution and high-quality images on mobile platforms. We propose several techniques to achieve this goal. First, we systematically examine the design choices of the network architecture to reduce model parameters and latency, while ensuring high-quality generation. Second, to further improve generation quality, we employ cross-architecture knowledge distillation from a much larger model, using a multi-level approach to guide the training of our model from scratch. Third, we enable a few-step generation by integrating adversarial guidance with knowledge distillation. For the first time, our model SnapGen, demonstrates the generation of 1024x1024 px images on a mobile device around 1.4 seconds. On ImageNet-1K, our model, with only 372M parameters, achieves an FID of 2.06 for 256x256 px generation. On T2I benchmarks (i.e., GenEval and DPG-Bench), our model with merely 379M parameters, surpasses large-scale models with billions of parameters at a significantly smaller size (e.g., 7x smaller than SDXL, 14x smaller than IF-XL).