paper-with-me

홈 › Papers

Multistep Distillation of Diffusion Models via Moment Matching

2024-06-06 · Tim Salimans, Thomas Mensink, Jonathan Heek, Emiel Hoogeboom

We present a new method for making diffusion models faster to sample. The method distills many-step diffusion models into few-step models by matching conditional expectations of the clean data given noisy data along the sampling trajectory. Our approach extends recently proposed one-step methods to the multi-step case, and provides a new perspective by interpreting these approaches in terms of moment matching. By using up to 8 sampling steps, we obtain distilled models that outperform not only their one-step versions but also their original many-step teacher models, obtaining new state-of-the-art results on the Imagenet dataset. We also show promising results on a large text-to-image model where we achieve fast generation of high resolution images directly in image space, without needing autoencoders or upsamplers.

📄 PDF Abstract BibTeX arXiv:2406.04103

Code (0)

등록된 구현이 없습니다.

Methods 이 논문이 사용한 방법론

Diffusion Diffusion models generate samples by gradually removing noise from a signal, and their training objective can be expressed as a reweighted variational lower-bound…

Similar Papers 제목 키워드 기반

Multi-student Diffusion Distillation for Better One-step Generators

2024-10-30 · Yanke Song, Jonathan Lorraine, Weili Nie, Karsten Kreis 외

Diffusion models achieve high-quality sample generation at the cost of a lengthy multistep inference procedure. To overcome this, diffusion distillation techniques produce student generators capable of matching or surpas…

Image Generation

Diffusion Fine-tuning with Rewarded Moment Matching Distillation

2026-06-29 · Alexis Jacq, Guillaume Couairon, Valentin De Bortoli, Quentin Berthet 외 arxiv

Distillation and Reinforcement Learning (RL) fine-tuning are the primary pillars of diffusion post-training. While traditionally studied in isolation, the interaction between these phases remains poorly understood, and i…

Reinforcement LearningWeather Forecasting

Beyond Single Tokens: Distilling Discrete Diffusion Models via Discrete MMD

2026-03-20 · Emiel Hoogeboom, David Ruhe, Jonathan Heek, Thomas Mensink 외 arxiv

It is currently difficult to distill discrete diffusion models. In contrast, continuous diffusion literature has many distillation approaches methods that can reduce sampling steps to a handful. Our method, Discrete Mome…

Multistep Consistency Models

2024-03-11 · Jonathan Heek, Emiel Hoogeboom, Tim Salimans

Diffusion models are relatively easy to train but require many steps to generate samples. Consistency models are far more difficult to train, but generate samples in a single step. In this paper we propose Multistep Cons…

Inductive Moment Matching

2025-03-10 · Linqi Zhou, Stefano Ermon, Jiaming Song

Diffusion models and Flow Matching generate high-quality samples but are slow at inference, and distilling them into few-step models often leads to instability and extensive tuning. To resolve these trade-offs, we propos…