Diffusion models have become a dominant paradigm for high-quality generative modeling, while post-training is essential for adapting them to diverse downstream applications. However, post-training of large diffusion models is still challenging due to the prohibitive memory footprints and slow training speed, which existing parameter-efficient fine-tuning methods only partially address. To overcome these limitations, we propose FourTune, an efficient post-training framework for diffusion models based on an end-to-end W4A4G4 paradigm. FourTune introduces a triple-branch hybrid pipeline that augments the standard LoRA architecture with a frozen numerical stabilizer to isolate quantization-sensitive outliers, enabling stable training under native 4-bit computation. In addition, FourTune employs hardware-efficient block-wise quantization and customized fused kernels to support efficient quantized backpropagation and reduce memory bandwidth overhead. Across customization, reinforcement learning, and distillation tasks, FourTune matches the quality of full-precision fine-tuning. On FLUX.1-dev (12B), FourTune reduces memory overhead by 2.25× and increases end-to-end training throughput by 2.27× compared to BF16 LoRA.
@article{xue2026fourtune,
title = {{FourTune}: Towards Fully 4-Bit Efficient Post-Training for Diffusion Models},
author = {Xue, Bowen and Min, Zihan and Li, Xingyang and Zhang, Zhekai and
Xi, Haocheng and Zhang, Lvmin and Agrawala, Maneesh and Zhu, Jun-Yan and
Han, Song and Lin, Yujun and Li, Muyang},
journal = {arXiv preprint arXiv:2607.05711},
year = {2026},
}
We thank MIT-IBM Watson AI Lab, MIT and Amazon Science Hub, MIT AI Hardware Program, National Science Foundation, Packard Foundation, Dell, LG, Hyundai, and Samsung for supporting this research. We thank NVIDIA for donating the DGX server.
