VDE：通过速度分解与估计实现整流流模型的免训练加速
VDE: Training-Free Accelerating Rectified Flow Model via Velocity Decomposition and Estimation
把整流流模型的速度场拆成平行/正交分量做输入自适应估计，替代静态特征缓存，可直接套到 Flux、Qwen-Image 等扩散/流匹配采样器上做免训练少步加速。
原文依据 VDE decomposes the model's velocity into components parallel and orthogonal to the input
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论尽管整流流模型在图像、视频与三维生成中取得了显著性能，其实际部署仍受推理速度慢的制约。此前的加速方法复用前序步骤的缓存特征，忽略了静态缓存与不断演化的输入之间日益增大的失配，导致输出保真度下降。本工作提出速度分解与估计（VDE），一种免训练的加速方法，将范式从「缓存并复用」转变为「分解并估计」。具体而言，VDE 将模型的速度分解为与输入平行和正交的分量，利用它们的时序可预测性与方向一致性进行精确的、输入自适应的估计。为防止误差累积，它通过完整前向传播周期性地锚定模型状态。在图像与视频生成任务上的大量实验表明，VDE 在视觉质量损失极小的前提下实现了显著加速。值得注意的是，VDE 将 Flux 加速 3.22 倍，并在 Qwen-Image 上取得 0.069 的感知相似度，相较最佳基线降低 52.2%。代码：https://github.com/Tan-Junwen/VDE
英文原文摘要 1181 字符
Though rectified flow models have achieved remarkable performance in image, video, and 3D generation, their practical deployments are challenged by slow inference speeds. Prior acceleration methods reuse cached features from previous steps, which neglects the growing mismatch between static caches and the evolving input, leading to reduced output fidelity. This work proposes Velocity Decomposition and Estimation (VDE), a training-free acceleration method that shifts the paradigm from caching-and-reusing to decomposing-and-estimating. Specifically, VDE decomposes the model's velocity into components parallel and orthogonal to the input, exploiting their temporal predictability and directional consistency for precise, input-adaptive estimation. To prevent error accumulation, it periodically anchors the model's state via full forward passes. Extensive experiments on image and video generation tasks demonstrate that VDE achieves substantial acceleration with minimal loss in visual quality. Notably, VDE accelerates Flux by 3.22x and achieves an LPIPS of 0.069 on Qwen-Image, outperforming the best baseline with a 52.2% reduction. Code: https://github.com/Tan-Junwen/VDE