This repository contains Nunchaku-quantized versions of Qwen-Image, designed to generate high-quality images from text prompts, advances in complex text rendering. It is optimized for efficient inference while maintaining minimal loss in performance.

Model Details

Model Description

Developed by: Nunchaku Team
Model type: text-to-image
License: apache-2.0
Quantized from model: Qwen-Image

Model Files

svdq-int4_r32-qwen-image.safetensors: SVDQuant quantized INT4 Qwen-Image model with rank 32. For users with non-Blackwell GPUs (pre-50-series).
svdq-int4_r128-qwen-image.safetensors: SVDQuant quantized INT4 Qwen-Image model with rank 128. For users with non-Blackwell GPUs (pre-50-series). It offers better quality than the rank 32 model, but it is slower.
svdq-fp4_r32-qwen-image.safetensors: SVDQuant quantized NVFP4 Qwen-Image model with rank 32. For users with Blackwell GPUs (50-series).
svdq-fp4_r128-qwen-image.safetensors: SVDQuant quantized NVFP4 Qwen-Image model with rank 128. For users with Blackwell GPUs (50-series). It offers better quality than the rank 32 model, but it is slower.

Model Sources

Inference Engine: nunchaku
Quantization Library: deepcompressor
Paper: SVDQuant: Absorbing Outliers by Low-Rank Components for 4-Bit Diffusion Models
Demo: svdquant.mit.edu

Usage

Diffusers Usage: See qwen-image.py.
ComfyUI Usage: Coming soon!

Performance

Citation

@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}
}