Theory of Consistency Diffusion Models: Distribution Estimation Meets Fast Sampling

Authors: Zehao Dou, Minshuo Chen, Mengdi Wang, Zhuoran Yang

ICML 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical This paper contributes towards the first statistical theory for consistency models, formulating their training as a distribution discrepancy minimization problem. Our analysis yields statistical estimation rates based on the Wasserstein distance for consistency models, matching those of vanilla diffusion models. Additionally, our results encompass the training of consistency models through both distillation and isolation methods, demystifying their underlying advantage.
Researcher Affiliation Academia 1Department of Statistics and Data Science, Yale University, New Haven, US 2Electrical and Computer Engineering, Princeton University, Princeton, US.
Pseudocode No The paper does not contain any pseudocode or clearly labeled algorithm blocks.
Open Source Code No The paper does not provide any statements or links regarding the availability of open-source code for the described methodology.
Open Datasets No The paper is theoretical and does not use or describe a specific dataset for experiments with public access information. It refers to 'the given dataset is {xj}j [n], which is assumed to be i.i.d sampled from pdata' but this is a theoretical notation, not a specific dataset that was used in experiments.
Dataset Splits No The paper is theoretical and does not report on experiments or data splits for training, validation, or testing.
Hardware Specification No The paper is theoretical and does not describe any hardware used for experiments.
Software Dependencies No The paper is theoretical and does not list any specific software dependencies with version numbers for reproducibility of experiments.
Experiment Setup No The paper is theoretical and does not describe any experimental setup details such as hyperparameters or training configurations.