自动调整生成步数，让一致性模型训练更快更稳。
Adaptive Discretization for Consistency Models
- 把离散化过程变成优化问题，用局部与全局一致性平衡训练稳定性。
- 在CIFAR-10和ImageNet上以极低开销实现更好生成效果。
- 适合想提升训练效率的扩散模型研究者，尤其适配高级变体。
一致性模型（CMs）在单步生成中表现优异，但现有方法依赖人工设计的离散化方案，需针对不同噪声调度和数据集反复调参。为此，我们提出统一的自适应离散化框架，将离散化步骤建模为优化问题。训练过程中，以局部一致性为优化目标，避免过度离散化；以全局一致性为约束，控制去噪误差，确保稳定性，并通过拉格朗日乘子权衡两者。基于此，我们采用高斯-牛顿法实现自适应离散化，命名为ADCMs。实验表明，ADCMs显著提升训练效率，在CIFAR-10和ImageNet上均以最小训练开销获得更优生成性能，且对更先进的扩散模型变体具备强适应性。代码已开源。
原文摘要 · Abstract (English)
Consistency Models (CMs) have shown promise for efficient one-step generation. However, most existing CMs rely on manually designed discretization schemes, which can cause repeated adjustments for different noise schedules and datasets. To address this, we propose a unified framework for the automatic and adaptive discretization of CMs, formulating it as an optimization problem with respect to the discretization step. Concretely, during the consistency training process, we propose using local consistency as the optimization objective to ensure trainability by avoiding excessive discretization, and taking global consistency as a constraint to ensure stability by controlling the denoising error in the training target. We establish the trade-off between local and global consistency with a Lagrange multiplier. Building on this framework, we achieve adaptive discretization for CMs using the Gauss-Newton method. We refer to our approach as ADCMs. Experiments demonstrate that ADCMs significantly improve the training efficiency of CMs, achieving superior generative performance with minimal training overhead on both CIFAR-10 and ImageNet. Moreover, ADCMs exhibit strong adaptability to more advanced DM variants. Code is available at https://github.com/rainstonee/ADCM.
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。