用副本交换方法实现扩散模型推理时的灵活控制。
CREPE: Controlling Diffusion with Replica Exchange
- 基于副本交换算法，推理时逐个生成样本并保持多样性。
- 支持在线优化或提前终止，烧灼期后样本多样性高。
- 适用于温度调节、奖励倾斜等任务，性能媲美传统方法。
推理阶段对扩散模型进行控制，旨在不重新训练的情况下引导输出满足新约束。以往方法多依赖启发式引导，或与序列蒙特卡洛（SMC）结合用于偏差校正。本文提出一种基于副本交换的灵活替代方案，称为CREPE（Controlling with REPlica Exchange）。与SMC不同，CREPE具备三个优势：（1）逐个生成粒子；（2）在烧灼期后保持高样本多样性；（3）支持在线优化或提前终止。我们在多种任务中验证其通用性，包括温度退火、奖励倾斜、模型组合及无分类器引导去偏，性能与先前的SMC方法相当。
原文摘要 · Abstract (English)
Inference-time control of diffusion models aims to steer model outputs to satisfy new constraints without retraining. Previous approaches have mostly relied on heuristic guidance or have been coupled with Sequential Monte Carlo (SMC) for bias correction. In this paper, we propose a flexible alternative based on replica exchange, an algorithm designed initially for sampling problems. We refer to this method as CREPE (Controlling with REPlica Exchange). Unlike SMC, CREPE: (1) generates particles sequentially, (2) maintains high diversity in the generated samples after a burn-in period, and (3) enables online refinement or early termination. We demonstrate its versatility across various tasks, including temperature annealing, reward-tilting, model composition and classifier-free guidance debiasing, with competitive performance compared to prior SMC methods.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。