metadata
license: mit
pipeline_tag: image-to-image
library_name: diffusers
REPA-E: Unlocking VAE for End-to-End Tuning of Latent Diffusion Transformers
[🌐 Project Page](https://end2end-diffusion.github.io) [📃 Paper](https://arxiv.org/abs/2504.10483) [🤗 Github](https://github.com/REPA-E/REPA-E)
REPA-E enables stable and effective joint training of both the VAE and the diffusion model, significantly accelerating training and improving generation quality. It achieves state-of-the-art FID scores on ImageNet 256×256. For detailed usage instructions, including environment setup, training, and evaluation, please refer to the project page and the GitHub repository.