一致性模型的多步采样理论：稳定性、误差界与噪声调度

View PDF HTML (experimental)

Abstract:Consistency models (CMs) have become a leading approach for generating high-quality samples in few steps. However, adding steps can improve or degrade sample quality in ways that are highly sensitive to the schedule and that existing theory does not fully explain. To provide accuracy guarantees and guide CM sampler design, we analyze multistep CM sampling as a composition of noising and approximate denoising operators. Under explicit, verifiable stability assumptions, we derive a non-asymptotic error bound that separates contraction of the initialization error from accumulation of approximation error. The bound assigns distinct roles to the schedule: large early noise levels drive contraction, while small late noise levels control the residual bias. As a corollary, we obtain explicit constants for strongly log-concave and semi-log-concave targets. We further establish a complementary guarantee whose assumptions, one-step accuracy and stability, can be estimated for a given trained model. Experiments show that the contraction and approximation profiles entering our bounds can be reliably measured and closely match the predicted functional forms. Together, these results provide a meaningful convergence theory for multi-step CMs and a practical route to sampler design.

Comments:	27 pages, 6 figures
Subjects:	Machine Learning (stat.ML); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Cite as:	arXiv:2610.03414 [stat.ML]
	(or arXiv:2610.03414v1 [stat.ML] for this version)
	https://doi.org/10.48550/arXiv.2610.03414 arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Alessio Spagnoletti [view email]
[v1] Fri, 2 Oct 2026 15:00:49 UTC (88 KB)