Abstract:We present a mechanism to steer the sampling diversity of denoising diffusion and flow matching models, allowing users to sample from a sharper or broader distribution than the training distribution. We build on the observation that these models leverage (learned) score functions of noisy data distributions for sampling and show that rescaling these allows one to effectively control a 'local' sampling temperature. Notably, this approach does not require any finetuning or alterations to training strategy, and can be applied to any off-the-shelf model and is compatible with both deterministic and stochastic samplers. We first validate our framework on toy 2D data, and then demonstrate its application for diffusion models trained across five disparate tasks -- image generation, pose estimation, depth prediction, robot manipulation, and protein design. We find that across these tasks, our approach allows sampling from sharper (or flatter) distributions, yielding performance gains e.g., depth prediction models benefit from sampling more likely depth estimates, whereas image generation models perform better when sampling a slightly flatter distribution.
| Comments: | Accepted at ICML 2026. Project page: this https URL |
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2510.01184 [cs.LG] |
| (or arXiv:2510.01184v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2510.01184 arXiv-issued DOI via DataCite |
From: Yu Wu [view email]
[v1]
Wed, 1 Oct 2025 17:59:51 UTC (27,028 KB)
[v2]
Fri, 22 May 2026 21:36:06 UTC (29,398 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。