arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2610.16448cs.CV
收藏

基于Tikhonov正则化贝塞尔表示的无训练扩散模型加速预测

Tikhonov-Stabilized Bezier Representation Forecasting for Training-free Diffusion Acceleration

School of Electronic and Computer Engineering, Peking University(北京大学电子与计算机工程学院);School of Informatics, Xiamen University(厦门大学信息学院);BNRist, Tsinghua University(清华大学BNRist)

Lei Zhu, Mujie Lin, Ruochong Zheng, Guangyi Wang, Hao Li, Peng Jin, Chang Liu, Jie Chen

AI总结:

该研究针对扩散模型推理成本高的问题,提出无训练的BeziCast框架,通过Tikhonov正则化贝塞尔轨迹预测实现加速,可将FLUX.1、HunyuanVideo分别最高加速4.79倍、4.11倍,且生成质量优于竞争基线。

摘要:

扩散模型,尤其是扩散Transformer(Diffusion Transformers),在图像和视频生成领域取得了优异的质量,但由于需要反复评估去噪器,推理阶段的成本仍然很高。特征缓存是一种便于部署的加速策略,它通过复用或预测中间表示,无需对生成器进行重新训练或修改采样器。然而,现有的“缓存后预测”方法通常依赖泰勒式有限差分外推,在较长的缓存间隔下容易变得不稳定,且通常预测局部模块的输出,其误差可能会在后续的去噪模块中累积。我们提出了BeziCast,一种无训练的扩散模型加速框架,它使用低阶贝塞尔(Bézier)轨迹来预测接近去噪器输出的表示。BeziCast通过Tikhonov正则化拟合来估计贝塞尔控制点,提供了一种平滑、容量可控的时间参数化,避免了显式的高阶导数估计,并将轨迹容量与缓存查询密度解耦。为了抑制激进的外推,我们进一步推导了等效预测权重,并引入了受凸包启发的防护机制,该机制可检测高风险预测并将其柔和地拉向保守的单纯形投影估计。在先进的图像和视频扩散模型上进行的广泛评估证明了BeziCast的有效性。特别是,BeziCast可将FLUX.1加速最高达4.79倍,将HunyuanVideo加速最高达4.11倍,同时相比竞争的加速基线,能保持显著更好的生成质量。

英文摘要:

Diffusion models, particularly Diffusion Transformers, achieve strong image and video generation quality but remain expensive at inference time due to repeated denoiser evaluations. Feature caching offers a deployment-friendly acceleration strategy by reusing or predicting intermediate representations without retraining the generator or modifying the sampler. However, existing cache-then-forecast methods often rely on Taylor-style finite-difference extrapolation, which can become unstable over longer cache intervals, and typically predict local module outputs whose errors may accumulate through subsequent denoiser blocks. We propose \textbf{BeziCast}, a training-free diffusion acceleration framework that forecasts output-proximal denoising representations using low-order B'ezier trajectories. BeziCast estimates B'ezier control points via Tikhonov-stabilized fitting, providing a smooth, capacity-controlled temporal parameterization that avoids explicit high-order derivative estimation and decouples trajectory capacity from cache-query density. To moderate aggressive extrapolation, we further derive equivalent forecasting weights and introduce a convex-hull-inspired guardrail that detects high-risk predictions and softly pulls them toward a conservative simplex-projected estimate. Extensive evaluations across advanced image and video diffusion models demonstrate the effectiveness of BeziCast. In particular, BeziCast accelerates FLUX.1 by up to (4.79\times) and HunyuanVideo by up to (4.11\times), while preserving substantially better generation quality than competing acceleration baselines.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑