{"ID":2853240,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2510.16983","arxiv_id":"2510.16983","title":"One-step Diffusion Models with Bregman Density Ratio Matching","abstract":"Diffusion and flow models achieve high generative quality but remain computationally expensive due to slow multi-step sampling. Distillation methods accelerate them by training fast student generators, yet most existing objectives lack a unified theoretical foundation. In this work, we propose Di-Bregman, a compact framework that formulates diffusion distillation as Bregman divergence-based density-ratio matching. This convex-analytic view connects several existing objectives through a common lens. Experiments on CIFAR-10 and text-to-image generation demonstrate that Di-Bregman achieves improved one-step FID over reverse-KL distillation and maintains high visual fidelity compared to the teacher model. Our results highlight Bregman density-ratio matching as a practical and theoretically-grounded route toward efficient one-step diffusion generation.","short_abstract":"Diffusion and flow models achieve high generative quality but remain computationally expensive due to slow multi-step sampling. Distillation methods accelerate them by training fast student generators, yet most existing objectives lack a unified theoretical foundation. In this work, we propose Di-Bregman, a compact fra...","url_abs":"https://arxiv.org/abs/2510.16983","url_pdf":"https://arxiv.org/pdf/2510.16983v1","authors":"[\"Yuanzhi Zhu\",\"Eleftherios Tsonis\",\"Lucas Degeorge\",\"Vicky Kalogeiton\"]","published":"2025-10-19T20:00:54Z","proceeding":"cs.CV","tasks":"[\"cs.CV\",\"cs.AI\",\"cs.LG\"]","methods":"[\"Diffusion Model\"]","has_code":false}