RBF-Solver: A Multistep Sampler for Diffusion Probabilistic Models via Radial Basis Functions

Soochul Park, Yeon Ju Lee, SeongJin Yoon, [4pt] Jiyub Shin, Juhee Lee, Seongwoon Jo · 2026

Read RBF-Solver: A Multistep Sampler for Diffusion Probabilistic Models via Radial Basis Functions by Soochul Park, Yeon Ju Lee, SeongJin Yoon, [4pt] Jiyub Shin, Juhee Lee, Seongwoon Jo (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-13330.txt.

Description

Soochul Park and co-authors propose RBF-Solver, a faster sampler for diffusion probabilistic models. Sampling from these models gives high fidelity but costs a lot of compute. Polynomial multistep samplers speed things up, yet follow a fixed path with no room to tune. RBF-Solver interpolates model outputs with Gaussian radial basis functions that have learnable shape values. At first order it matches the Euler method, and at higher order it nears the Adams method. It beat polynomial samplers on CIFAR-10 and ImageNet tests at set function counts.

Catalog record

CREATORSoochul Park, Yeon Ju Lee, SeongJin Yoon, [4pt] Jiyub Shin, Juhee Lee, Seongwoon Jo
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE130413
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.13330

Browse the full library · About the Tower of Babel