Adaptive Moments are Surprisingly Effective for Plug-and-Play Diffusion Sampling

ICLR 2026 Conference Submission15226 Authors

Keywords: Guided Diffusion Sampling, Plug-and-Play Conditional Diffusion Sampling, Adaptive Moment Estimation

TL;DR: Adaptive moments are surprisingly effective for plug-and-play diffusion sampling.

Abstract: Guided diffusion sampling relies on approximating intractable likelihood scores, which introduces significant noise into the sampling dynamics. We propose using adaptive moment estimation to stabilize these noisy likelihood scores during sampling. Despite its simplicity, our approach achieves state-of-the-art results on image restoration and class-conditional generation tasks, outperforming more complicated methods, which are often computationally more expensive. We provide empirical analysis of our method on both synthetic and real data, demonstrating that mitigating gradient noise through adaptive moments offers an effective way to improve alignment.

Primary Area: generative models

Submission Number: 15226