提出SteinDiff框架，解决扩散模型大步长推理中的稳定性问题。
Mitigating the Contractivity Trap in Diffusion ODEs via Stein Stabilization
- 用无参考样本的Stein修正机制，动态调整大步长求解器更新。
- 在大步长推理下显著减少伪影，提升生成质量，验证了稳定性提升。
- 适合追求高效推理且关注生成稳定性的研究者与应用开发者。
扩散模型在通过确定性概率流常微分方程（PF-ODE）进行大步长推理时面临根本性矛盾，即‘收缩陷阱’：高效推理依赖大步长，但激进步长与高表达能力去噪器会破坏基于收缩性的误差抑制稳定性保证。为此，我们提出SteinDiff，一种无需参考样本的推理时稳定化框架，采用基于Stein的修正机制。具体而言，SteinDiff引入几何感知的残差修正机制，在不重训练的前提下正则化大步长求解器更新。我们推导出可用于步间调整的闭式Stein修正系数，实现对局部数据几何的无参考适应。此外，我们在分布偏移下建立了受得分控制的扰动界，并提供了对EDM型参数化的互补Stein视角。大量实验表明，SteinDiff有效缓解严重伪影，提升各类大步长推理场景下的生成质量。
原文摘要 · Abstract (English)
A fundamental tension exists in the large-step inference of diffusion models via their deterministic probability flow ordinary differential equation (PF-ODE) trajectories, which we identify as the contractivity trap: efficient inference favors large step sizes, while aggressive steps and highly expressive denoisers can undermine contraction-based stability certificates for error suppression. To address this, we propose SteinDiff, a step-wise inference-time stabilization framework that employs Stein-derived corrections without requiring reference samples. Specifically, SteinDiff introduces a geometry-aware residual correction mechanism that regularizes large-step solver updates without retraining. To this end, we derive a closed-form Stein correction coefficient for step-wise solver adjustment, enabling reference-free adaptation to local data geometry. We further establish a score-controlled perturbation bound under distributional shifts and provide a complementary Stein perspective on EDM-style parameterizations. Extensive experiments demonstrate that SteinDiff mitigates severe artifacts and improves generative quality across large-step inference settings.
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。