SVDQunat: Absorbing Outliers by Low-Rank Components for 4-Bit Diffusion Models
Paper • 2411.05007 • Published • 25
This is a "fork of the community maintained nunchaku (Vitoom Nunchaku)", farther extended that gives you full control over Nunchaku Flux.2 Klein".
For more info: ComfyUI-Nunchaku-Klein
Wheel-repacking scripts: vitoom-nunchaku-extended
@inproceedings{
li2024svdquant,
title={SVDQuant: Absorbing Outliers by Low-Rank Components for 4-Bit Diffusion Models},
author={Li*, Muyang and Lin*, Yujun and Zhang*, Zhekai and Cai, Tianle and Li, Xiuyu and Guo, Junxian and Xie, Enze and Meng, Chenlin and Zhu, Jun-Yan and Han, Song},
booktitle={The Thirteenth International Conference on Learning Representations},
year={2025}
}
Base model
nunchaku-ai/nunchaku