Scale-RAE: Scaling Text-to-Image Diffusion Transformers with Representation Autoencoders

Official model weights for the paper Scaling Text-to-Image Diffusion Transformers with Representation Autoencoders.

Representation Autoencoders (RAEs) enable diffusion modeling in high-dimensional semantic latent spaces. Scale-RAE scales this framework to large-scale, freeform text-to-image generation. RAEs consistently outperform traditional VAEs during pretraining across various model scales, offering faster convergence and better generation quality.

• Project Page: https://rae-dit.github.io/scale-rae/
• GitHub Repository: https://github.com/ZitengWangNYU/Scale-RAE
• Paper: https://arxiv.org/abs/2601.16208

Usage

For full text-to-image generation using Scale-RAE, please follow the installation and inference instructions in the official repository.

Citation

@article{scale-rae-2026,
 title={Scaling Text-to-Image Diffusion Transformers with Representation Autoencoders},
 author={Shengbang Tong and Boyang Zheng and Ziteng Wang and Bingda Tang and Nanye Ma and Ellis Brown and Jihan Yang and Rob Fergus and Yann LeCun and Saining Xie},
 journal={arXiv preprint arXiv:2601.16208},
 year={2026}
}