给定计算预算，自动找最优缓存策略，提升扩散模型生成质量
Budget-Constrained Step-Level Diffusion Caching
- 先设定计算预算，再搜索最优缓存步骤组合
- 在FLUX.1-dev和Wan2.1上优于传统启发式缓存方法
- 离线搜索快速生成策略，推理时无额外开销
步级缓存通过利用去噪过程中的时间冗余加速扩散模型。现有方法采用阈值启发式决定每步是否缓存，未直接优化最终输出质量，导致推理延迟随输入波动，部署时难以控制。本文提出BudCache，反其道而行之：预先设定计算预算，搜索能最好保留最终输出质量的缓存策略。为应对步选择的组合复杂性，结合模拟退火与确定性爬山算法。该离线搜索在几分钟内完成，推理阶段无在线搜索或阈值开销。当预算极紧时，进一步引入缓存感知的时间离散对齐，以减少缓存引起的轨迹偏差。在FLUX.1-dev和Wan2.1上的实验表明，相同推理预算下，BudCache生成质量优于启发式缓存基线。
原文摘要 · Abstract (English)
Step-level caching accelerates diffusion models by exploiting temporal redundancy across denoising steps. Existing methods make per-step cache decisions using threshold-based heuristics, without directly optimizing for final output quality. As a result, their inference latency varies across inputs and is difficult to control at deployment. In this work, we propose BudCache, which inverts this formulation: rather than letting per-step error thresholds dictate the runtime cost, we fix the compute budget in advance and search for the cache policy that best preserves the final output. To tackle the combinatorial complexity of step selection, we combine Simulated Annealing with deterministic Hill Climbing. This offline search identifies high-quality cache policies within minutes and introduces no online search or thresholding overhead during inference. When the compute budget is very tight, we further introduce cache-aware schedule alignment, which adapts the time discretization to the selected cache policy to reduce cache-induced trajectory mismatch. Experiments on FLUX.1-dev and Wan2.1 show that BudCache achieves better generation quality than heuristic caching baselines under the same inference budgets. Code is available at https://github.com/Westlake-AGI-Lab/BudCache
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。