English

Diffusion-SDPO: Safeguarded Direct Preference Optimization for Diffusion Models

Computer Vision and Pattern Recognition 2025-12-03 v2

Abstract

Text-to-image diffusion models deliver high-quality images, yet aligning them with human preferences remains challenging. We revisit diffusion-based Direct Preference Optimization (DPO) for these models and identify a critical pathology: enlarging the preference margin does not necessarily improve generation quality. In particular, the standard Diffusion-DPO objective can increase the reconstruction error of both winner and loser branches. Consequently, degradation of the less-preferred outputs can become sufficiently severe that the preferred branch is also adversely affected even as the margin grows. To address this, we introduce Diffusion-SDPO, a safeguarded update rule that preserves the winner by adaptively scaling the loser gradient according to its alignment with the winner gradient. A first-order analysis yields a closed-form scaling coefficient that guarantees the error of the preferred output is non-increasing at each optimization step. Our method is simple, model-agnostic, broadly compatible with existing DPO-style alignment frameworks and adds only marginal computational overhead. Across standard text-to-image benchmarks, Diffusion-SDPO delivers consistent gains over preference-learning baselines on automated preference, aesthetic, and prompt alignment metrics. Code is publicly available at https://github.com/AIDC-AI/Diffusion-SDPO.

Keywords

Cite

@article{arxiv.2511.03317,
  title  = {Diffusion-SDPO: Safeguarded Direct Preference Optimization for Diffusion Models},
  author = {Minghao Fu and Guo-Hua Wang and Tianyu Cui and Qing-Guo Chen and Zhao Xu and Weihua Luo and Kaifu Zhang},
  journal= {arXiv preprint arXiv:2511.03317},
  year   = {2025}
}

Comments

The code is publicly available at https://github.com/AIDC-AI/Diffusion-SDPO