analiza aiqbench
AREX presents a novel, training-free sampling method for flow matching models that decomposes dynamics into affine and residual components to enhance few-step generation fidelity. By analytically handling the linear portion of the velocity field, it improves sample quality in image and text-to-image tasks without requiring model retraining. This advancement offers a practical efficiency boost for deploying pretrained generative AI systems with reduced inference steps.
AREX: Affine-Residual Exponential Integrator for Few-Step Sampling in Flow Matching
arXiv:2610.03483v1 Announce Type: cross Abstract: We introduce AREX, a training-free sampler for pretrained flow matching models that uses the target mean and covariance to capture an analytically tractable part of the sampling dynamics. We show that the velocity field of the moment-matched Gaussian target is the $L^2$-optimal affine approximation to the marginal velocity field. This motivates decomposition of the learned dynamics into an affine component over the whole sampling path, determined by the first two target moments, and a neural residual term. AREX keeps the affine component and integrates it using an explicit matrix-valued propagator. In turn, we only require to integrate over the residual term. This differs from scalar exponential integrators, which analytically handle only isotropic linear dynamics. Across image and text-to-image generation tasks, AREX consistently improves sample fidelity in the few-step sampling regime without retraining the underlying model.
Abstract:We introduce AREX, a training-free sampler for pretrained flow matching models that uses the target mean and covariance to capture an analytically tractable part of the sampling dynamics. We show that the velocity field of the moment-matched Gaussian target is the $L^2$-optimal affine approximation to the marginal velocity field. This motivates decomposition of the learned dynamics into an affine component over the whole sampling path, determined by the first two target moments, and a neural residual term. AREX keeps the affine component and integrates it using an explicit matrix-valued propagator. In turn, we only require to integrate over the residual term. This differs from scalar exponential integrators, which analytically handle only isotropic linear dynamics. Across image and text-to-image generation tasks, AREX consistently improves sample fidelity in the few-step sampling regime without retraining the underlying model.
Machine Learning (stat.ML); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
arXiv:2610.03483 [stat.ML]
(or arXiv:2610.03483v1 [stat.ML] for this version)
https://doi.org/10.48550/arXiv.2610.03483 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
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.
Text integral sub licența deschisă a publicației · 235 cuvinte
Subiecte conexe
TraceSampler: Making Sampling Errors Predictable and Correctable for Accelerated Generative Sampling
arXiv · 15 oct. 2026
$\Delta$ Flow Matching for Robot Foundation Models
arXiv · 14 oct. 2026
Learning to Solve Generative ODEs Beyond the Linear Span
arXiv · 9 iun. 2026
GraspMeanFlow: SE(3)-Equivariant MeanFlow for Few-Step 6-DoF Grasp Generation
arXiv · 5 aug. 2026