LiteVAE: Lightweight and Efficient Variational Autoencoders for Latent Diffusion Models

TL;DR

LiteVAE introduces a wavelet-based autoencoder for latent diffusion models that significantly reduces computational costs while maintaining or improving image quality, enabling faster training and lower memory usage.

Contribution

The paper presents LiteVAE, a novel wavelet-based autoencoder design that enhances scalability and efficiency in latent diffusion models without sacrificing output quality.

Findings

LiteVAE achieves comparable quality to standard VAEs with six times fewer encoder parameters.

LiteVAE's larger model outperforms existing VAEs across multiple quality metrics.

The proposed enhancements improve training dynamics and reconstruction quality.

Abstract

Advances in latent diffusion models (LDMs) have revolutionized high-resolution image generation, but the design space of the autoencoder that is central to these systems remains underexplored. In this paper, we introduce LiteVAE, a new autoencoder design for LDMs, which leverages the 2D discrete wavelet transform to enhance scalability and computational efficiency over standard variational autoencoders (VAEs) with no sacrifice in output quality. We investigate the training methodologies and the decoder architecture of LiteVAE and propose several enhancements that improve the training dynamics and reconstruction quality. Our base LiteVAE model matches the quality of the established VAEs in current LDMs with a six-fold reduction in encoder parameters, leading to faster training and lower GPU memory requirements, while our larger model outperforms VAEs of comparable complexity across all evaluated metrics (rFID, LPIPS, PSNR, and SSIM).