REPA-E/e2e-sdvae
<h1 align="center"> REPA-E: Unlocking VAE for End-to-End Tuning of Latent Diffusion Transformers </h1>
<p align="center"> <a href="https://scholar.google.com.au/citations?user=GQzvqS4AAAAJ" target="blank">Xingjian Leng</a><sup>1*</sup>   <b>·</b>   <a href="https://1jsingh.github.io/" target="blank">Jaskirat Singh</a><sup>1*</sup>   <b>·</b>   <a href="https://hou-yz.github.io/" target="blank">Yunzhong Hou</a><sup>1</sup>   <b>·</b>   <a href="https://people.csiro.au/X/Z/Zhenchang-Xing/" target="blank">Zhenchang Xing</a><sup>2</sup>  <b>·</b>   <a href="https://www.sainingxie.com/" target="blank">Saining Xie</a><sup>3</sup>  <b>·</b>   <a href="https://zheng-lab-anu.github.io/" target="blank">Liang Zheng</a><sup>1</sup>  </p>
<p align="center"> <sup>1</sup> Australian National University   <sup>2</sup>Data61-CSIRO   <sup>3</sup>New York University   <br> <sub><sup>*</sup>Project Leads </sub> </p>
<p align="center"> <a href="https://End2End-Diffusion.github.io">🌐 Project Page</a>   <a href="https://huggingface.co/REPA-E">🤗 Models</a>   <a href="https://arxiv.org/abs/2504.10483">📃 Paper</a>   <br> <!-- <a href="https://paperswithcode.com/sota/image-generation-on-imagenet-256x256?p=repa-e-unlocking-vae-for-end-to-end-tuning-of"><img src="https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/repa-e-unlocking-vae-for-end-to-end-tuning-of/image-generation-on-imagenet-256x256" alt="PWC"></a> --> </p>
<p align="center"> <img src="https://github.com/End2End-Diffusion/REPA-E/raw/main/assets/vis-examples.jpg" width="100%" alt="teaser"> </p>
We address a fundamental question: *Can latent diffusion models and their VAE tokenizer be trained end-to-end? While training both components jointly with standard diffusion loss is observed to be ineffective — often degrading final performance — we show that this limitation can be overcome using a simple representation-alignment (REPA) loss. Our proposed method, REPA-E*, enables stable and effective joint training of both the VAE and the diffusion model.
<p align="center"> <img src="https://github.com/End2End-Diffusion/REPA-E/raw/main/assets/overview.jpg" width="100%" alt="teaser"> </p>
REPA-E significantly accelerates training — achieving over 17× speedup compared to REPA and 45× over the vanilla training recipe. Interestingly, end-to-end tuning also improves the VAE itself: the resulting E2E-VAE provides better latent structure and serves as a drop-in replacement for existing VAEs (e.g., SD-VAE), improving convergence and generation quality across diverse LDM architectures. Our method achieves state-of-the-art FID scores on ImageNet 256×256: 1.12 with CFG and 1.69 without CFG.
Usage and Training
Please refer our Github Repo for detailed notes on end-to-end training and inference using REPA-E.
📚 Citation
@article{leng2025repae,
title={REPA-E: Unlocking VAE for End-to-End Tuning with Latent Diffusion Transformers},
author={Xingjian Leng and Jaskirat Singh and Yunzhong Hou and Zhenchang Xing and Saining Xie and Liang Zheng},
year={2025},
journal={arXiv preprint arXiv:2504.10483},
}