mlsanityData to 14 Oct 2026

One-step Diffusion Models with Bregman Density Ratio Matching

arXiv 2510.16983 ↗First posted 19 Oct 2025cs.CV cs.AI cs.LGcited by 1

On the map: Diffusion models → Diffusion foundations and sampling

Abstract (arXiv)

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.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Diffusion foundations and sampling, one of 14 directions in Diffusion models. Builds on 42 of its 54 references that are in the map. First posted on arXiv 19 Oct 2025.

Builds on

42 of 54 references.

Closest ancestors · 24

Same direction, newest first

  1. 2025Soft-Di[M]O: Improving One-Step Discrete Image Generation with Soft Embeddings26 Sep
  2. 2025Uni-Instruct: One-step Diffusion Model through Unified Diffusion Divergence Instruction27 May
  3. 2025Few-Step Diffusion via Score identity Distillation19 May
  4. 2025One-step Diffusion Models with f-Divergence Distribution Matching21 Feb
  5. 2024One-Step Diffusion Distillation through Score Implicit Matching22 Oct
  6. 2024Adversarial Score identity Distillation: Rapidly Surpassing the Teacher in One Step19 Oct
  7. 2024One Step Diffusion via Shortcut Models16 Oct
  8. 2024SwiftBrush v2: Make Your One-step Diffusion Model Better Than Its Teacher26 Aug
  9. 2024SlimFlow: Training Smaller One-Step Diffusion Models with Rectified Flow17 Jul
  10. 2024Multistep Distillation of Diffusion Models via Moment Matching6 Jun
  11. 2024Don't drop your samples! Coherence-aware training benefits Conditional diffusion30 May
  12. 2024EM Distillation for One-step Diffusion Models27 May
12 more
  1. 2024Improved Distribution Matching Distillation for Fast Image Synthesis23 May
  2. 2024Score identity Distillation: Exponentially Fast Distillation of Pretrained Diffusion Models for One-Step Generation5 Apr
  3. 2023One-step Diffusion with Distribution Matching Distillation30 Nov
  4. 2023Consistency Trajectory Models: Learning Probability Flow ODE Trajectory of Diffusion1 Oct
  5. 2023BOOT: Data-free Distillation of Denoising Diffusion Models with Bootstrapping8 Jun
  6. 2023Diff-Instruct: A Universal Approach for Transferring Knowledge From Pre-trained Diffusion Models29 May
  7. 2022Scalable Diffusion Models with Transformers19 Dec
  8. 2022On Distillation of Guided Diffusion Models6 Oct
  9. 2022Elucidating the Design Space of Diffusion-Based Generative Models1 Jun
  10. 2021High-Resolution Image Synthesis with Latent Diffusion Models20 Dec
  11. 2021Knowledge Distillation in Iterative Generative Models for Improved Sampling Speed7 Jan
  12. 2020Score-Based Generative Modeling through Stochastic Differential Equations26 Nov

Other references · 18

  1. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  2. 2022LAION-5B: An open large-scale dataset for training next generation image-text modelscited by 4,152
  3. 2022DreamFusion: Text-to-3D using 2D Diffusioncited by 3,025
  4. 2016Improved Techniques for Training GANscited by 2,970
  5. 2023ProlificDreamer: High-Fidelity and Diverse Text-to-3D Generation with Variational Score Distillationcited by 1,219
  6. 2025Mean Flows for One-step Generative Modelingcited by 561
  7. 2024MACE: Mass Concept Erasure in Diffusion Modelscited by 307
  8. 2022Rectified Flow: A Marginal Preserving Approach to Optimal Transportcited by 283
  9. 2023UFOGen: You Forward Once Large Scale Text-to-Image Generation via Diffusion GANscited by 166
  10. 2023SwiftBrush: One-Step Text-to-Image Diffusion Model with Variational Score Distillationcited by 110
  11. 2024Flow map matching with stochastic interpolants: A mathematical framework for consistency modelscited by 109
  12. 2025How to build a consistency model: Learning flow maps via self-distillationcited by 103
6 more
  1. 2024PeRFlow: Piecewise Rectified Flow as Universal Plug-and-Play Acceleratorcited by 97
  2. 2024AKiRa: Augmentation Kit on Rays for optical video generationcited by 18
  3. 2025Ultra-Fast Language Generation via Discrete Diffusion Divergence Instructcited by 16
  4. 2025Preference Optimization by Estimating the Ratio of the Data Distributioncited by 9
  5. 2025Pulp Motion: Framing-aware multimodal camera and human motion generationcited by 8
  6. 2025Training-Free Synthetic Data Generation with Dual IP-Adapter Guidancecited by 3

Built on by

Cited by year

Citing papers per year: 2025: 1.

1 paper.

  1. 2025TS-DP: Reinforcement Speculative Decoding For Temporal Adaptive Diffusion Policy Accelerationcited by 6

Similar, not cited

Close in wording, no citation either way

  1. 2026Consistent Distribution Matching for Data-Free Diffusion Distillation
  2. 2024Diffusion Models Are Innate One-Step Generators
  3. 2025Revisiting Diffusion Models: From Generative Pre-training to One-Step Generation
  4. 2024Simple and Fast Distillation of Diffusion Models
  5. 2024SFDDM: Single-fold Distillation for Diffusion models
  6. 2025Scale-wise Distillation of Diffusion Models
  7. 20261.x-Distill: Breaking the Diversity, Quality, and Efficiency Barrier in Distribution Matching Distillation
  8. 2025Learning Few-Step Diffusion Models by Trajectory Distribution Matching