aiqbench analysis
This paper introduces Slowly Annealed Langevin Dynamics (SALD) and its Velocity-Aware variant (VA-SALD) to improve training-free guided generation in pretrained score-based models. By establishing non-asymptotic convergence guarantees through KL differential inequalities, the authors provide a principled framework that clarifies how time slowdown corrects guidance bias in diffusion models. The work offers significant theoretical insights for practitioners seeking to enhance generative model performance without additional training.
Slowly Annealed Langevin Dynamics: Theory and Applications to Training-Free Guided Generation
arXiv:2605.07950v2 Announce Type: replace Abstract: We study Slowly Annealed Langevin Dynamics (SALD), a sampler for tracking a path of moving target distributions and approximating the terminal target through time slowdown. We establish non-asymptotic convergence guarantees via a KL differential inequality, showing that slowdown improves tracking through contraction of intermediate targets and the complexity of the path. Motivated by training-free guided generation with pretrained score-based generative models, we further introduce Velocity-Aware SALD (VA-SALD), which explicitly incorporates the underlying marginal distributions of the pretrained model and uses slowdown to correct the additional deviation induced by guidance. This yields a principled framework for training-free guided generation for diffusion-based and related generative model families, together with convergence guarantees that clarify the roles of intermediate functional inequalities and guidance bias. Code is available at https://github.com/anitan0925/sald.
Abstract:We study Slowly Annealed Langevin Dynamics (SALD), a sampler for tracking a path of moving target distributions and approximating the terminal target through time slowdown. We establish non-asymptotic convergence guarantees via a KL differential inequality, showing that slowdown improves tracking through contraction of intermediate targets and the complexity of the path. Motivated by training-free guided generation with pretrained score-based generative models, we further introduce Velocity-Aware SALD (VA-SALD), which explicitly incorporates the underlying marginal distributions of the pretrained model and uses slowdown to correct the additional deviation induced by guidance. This yields a principled framework for training-free guided generation for diffusion-based and related generative model families, together with convergence guarantees that clarify the roles of intermediate functional inequalities and guidance bias. Code is available at this https URL.
Machine Learning (cs.LG)
arXiv:2605.07950 [cs.LG]
(or arXiv:2605.07950v2 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2605.07950 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 219 words
