动态调整梯度权重，用更少步骤实现高质量图像生成。
DyWeight: Dynamic Gradient Weighting for Few-Step Diffusion Sampling
- 提出动态梯度加权机制，自适应融合历史梯度信息。
- 在多个数据集上仅用少量采样步数即达顶尖生成质量。
- 适合追求高效扩散模型推理的开发者与研究者使用。
扩散模型在多模态生成任务中表现卓越，但其采样过程因需数百次函数求值而效率低下。尽管多步常微分方程求解器通过重用历史梯度提升了效率，现有方法依赖人工设计的系数，无法适应扩散采样中非平稳的动力学特性。为此，我们提出动态梯度加权（DyWeight），一种轻量级、基于学习的多步求解器，引入简化的隐式耦合范式。通过放宽经典数值约束，DyWeight 学习无约束的时间可变参数，自适应聚合历史梯度，并内在缩放有效步长。该隐式时间校准使求解器数值轨迹精准匹配模型内部去噪动力学，即使在大步长下也能保持稳定。在 CIFAR-10、FFHQ、AFHQv2、ImageNet64、LSUN-Bedroom、Stable Diffusion 和 FLUX.1-dev 上的大量实验表明，DyWeight 在显著减少函数求值次数的同时，实现了更优的视觉保真度和稳定性，成为高效扩散求解器的新基准。代码已公开于 https://github.com/Westlake-AGI-Lab/DyWeight。
原文摘要 · Abstract (English)
Diffusion Models (DMs) have achieved state-of-the-art generative performance across multiple modalities, yet their sampling process remains prohibitively slow due to the need for hundreds of function evaluations. Recent progress in multi-step ODE solvers has greatly improved efficiency by reusing historical gradients, but existing methods rely on handcrafted coefficients that fail to adapt to the non-stationary dynamics of diffusion sampling. To address this limitation, we propose Dynamic Gradient Weighting (DyWeight), a lightweight, learning-based multi-step solver that introduces a streamlined implicit coupling paradigm. By relaxing classical numerical constraints, DyWeight learns unconstrained time-varying parameters that adaptively aggregate historical gradients while intrinsically scaling the effective step size. This implicit time calibration accurately aligns the solver's numerical trajectory with the model's internal denoising dynamics under large integration steps, avoiding complex decoupled parameterizations and optimizations. Extensive experiments on CIFAR-10, FFHQ, AFHQv2, ImageNet64, LSUN-Bedroom, Stable Diffusion and FLUX.1-dev demonstrate that DyWeight achieves superior visual fidelity and stability with significantly fewer function evaluations, establishing a new state-of-the-art among efficient diffusion solvers. Code is available at https://github.com/Westlake-AGI-Lab/DyWeight
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。