EpaCache：面向扩散式视觉生成的误差传播感知缓存技术
EpaCache: Error-Propagation-Aware Caching for Accelerating Diffusion-Based Visual Generation
查看机构详情
- JIIOV Technology（极目科技）
机构由 AI 辅助整理，请以论文原文为准。
浏览论文内容
中文总结 AI 辅助
该研究针对扩散式视觉生成模型推理成本高的问题，提出无训练的误差传播感知缓存策略EpaCache，在图像、视频合成任务上均实现了延迟与保真度的更优权衡。
中文摘要 AI 辅助
基于扩散的视觉生成模型可实现高质量图像与视频合成，但由于顺序采样器需反复评估大型网络，推理成本极高。现有的基于缓存的方法通过复用相邻时间步的中间计算来降低推理延迟，不过其缓存控制器主要依赖局部时间变化，未考虑缓存复用在轨迹层面的影响。本文提出无训练的缓存策略EpaCache（Error-Propagation-Aware Cache），可自适应地将复用预算分配给对下游影响较小的时间步。在图像与视频合成模型上的实验表明，EpaCache在延迟-保真度权衡上始终优于现有缓存方法：在FLUX.1-dev上，EpaCache的推理时间从11.7秒降至11.3秒，PSNR从21.4提升至22.8，在延迟和保真度上均优于现有最优缓存方法；在HunyuanVideo上，EpaCache相比未缓存推理实现2.63倍加速，且在匹配延迟下相比现有最优方法将SSIM从0.891提升至0.905。
英文摘要
Diffusion-based visual generative models deliver strong image and video synthesis quality but incur high inference costs because sequential samplers repeatedly evaluate large networks. Caching-based methods reduce inference latency by reusing intermediate computations across adjacent timesteps. However, existing cache controllers rely primarily on local temporal variation and overlook the trajectory-level consequences of cache reuse. We introduce Error-Propagation-Aware Cache (EpaCache), a training-free caching policy that adaptively allocates the reuse budget on timesteps with lower downstream impact. Experiments on image and video synthesis models demonstrate that EpaCache consistently improves the latency--fidelity trade-off over existing caching methods. On FLUX.1-dev, EpaCache outperforms the prior state-of-the-art caching method in both latency and fidelity, reducing inference time from $11.7$ s to $11.3$ s while improving PSNR from $21.4$ to $22.8$. On HunyuanVideo, EpaCache achieves a $2.63\times$ speedup over uncached inference and improves SSIM from $0.891$ to $0.905$ over the prior state-of-the-art method at matched latency.
