Abstract:We study the problem of training diffusion models to sample from a distribution with a given unnormalized density or energy function. We benchmark several diffusion-structured inference methods, including simulation-based variational approaches and off-policy methods (continuous generative flow networks). Our results shed light on the relative advantages of existing algorithms while bringing into question some claims from past work. We also propose a novel exploration strategy for off-policy methods, based on local search in the target space with the use of a replay buffer, and show that it improves the quality of samples on a variety of target distributions. Our code for the sampling methods and benchmarks studied is made public at this https URL as a base for future work on diffusion models for amortized inference.
From: Esmeralda Whitammer [view email]
[v1]
Wed, 7 Feb 2024 18:51:49 UTC (9,161 KB)
[v2]
Tue, 13 Feb 2024 16:32:09 UTC (9,161 KB)
[v3]
Sun, 26 May 2024 18:06:40 UTC (10,202 KB)
[v4]
Mon, 13 Jan 2025 09:56:11 UTC (10,208 KB)
[v5]
Fri, 28 Aug 2026 06:45:23 UTC (2,205 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。