REAR: Rethinking Visual Autoregressive Models via Generator-Tokenizer Consistency Regularization
Paper
•
2510.04450
•
Published
This repository provides model checkpoints for reAR introduced in the paper:
“REAR: Rethinking Visual Autoregressive Models via Generator-Tokenizer Consistency Regularization.” (arXiv:2510.04450)
Reference: arXiv abstract/HTML page.
This repo contains the following weight files:
reAR-S.safetensorsreAR-B.safetensorsreAR-L.safetensorsAll training / inference code and detailed instructions are maintained on GitHub:
This Hugging Face repo is mainly for hosting the released checkpoints.
If you use these checkpoints, please cite:
@article{he2025rear,
title = {REAR: Rethinking Visual Autoregressive Models via Generator-Tokenizer Consistency Regularization},
author = {He, Qiyuan and Li, Yicong and Ye, Haotian and Wang, Jinghao and Liao, Xinyao and Heng, Pheng-Ann and Ermon, Stefano and Zou, James and Yao, Angela},
journal = {arXiv preprint arXiv:2510.04450},
year = {2025}
}