Towards a mathematical theory for consistency training in diffusion models

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Li, Gen, Huang, Zhihan, Wei, Yuting
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866929241164087296
author Li, Gen
Huang, Zhihan
Wei, Yuting
author_facet Li, Gen
Huang, Zhihan
Wei, Yuting
contents Consistency models, which were proposed to mitigate the high computational overhead during the sampling phase of diffusion models, facilitate single-step sampling while attaining state-of-the-art empirical performance. When integrated into the training phase, consistency models attempt to train a sequence of consistency functions capable of mapping any point at any time step of the diffusion process to its starting point. Despite the empirical success, a comprehensive theoretical understanding of consistency training remains elusive. This paper takes a first step towards establishing theoretical underpinnings for consistency models. We demonstrate that, in order to generate samples within $\varepsilon$ proximity to the target in distribution (measured by some Wasserstein metric), it suffices for the number of steps in consistency learning to exceed the order of $d^{5/2}/\varepsilon$, with $d$ the data dimension. Our theory offers rigorous insights into the validity and efficacy of consistency models, illuminating their utility in downstream inference tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2402_07802
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Towards a mathematical theory for consistency training in diffusion models
Li, Gen
Huang, Zhihan
Wei, Yuting
Machine Learning
Information Theory
Statistics Theory
Consistency models, which were proposed to mitigate the high computational overhead during the sampling phase of diffusion models, facilitate single-step sampling while attaining state-of-the-art empirical performance. When integrated into the training phase, consistency models attempt to train a sequence of consistency functions capable of mapping any point at any time step of the diffusion process to its starting point. Despite the empirical success, a comprehensive theoretical understanding of consistency training remains elusive. This paper takes a first step towards establishing theoretical underpinnings for consistency models. We demonstrate that, in order to generate samples within $\varepsilon$ proximity to the target in distribution (measured by some Wasserstein metric), it suffices for the number of steps in consistency learning to exceed the order of $d^{5/2}/\varepsilon$, with $d$ the data dimension. Our theory offers rigorous insights into the validity and efficacy of consistency models, illuminating their utility in downstream inference tasks.
title Towards a mathematical theory for consistency training in diffusion models
topic Machine Learning
Information Theory
Statistics Theory
url https://arxiv.org/abs/2402.07802