BAG动态调度扩散模型缓存，兼顾预算全局感知与实例自适应。
BAG: Budget-Aware Gating for Diffusion Caching
- 用轻量门控网络动态决定每步是否复用缓存特征
- 在多模型多分辨率下实现速度提升10%-30%且稳定
- 适合需灵活适配不同计算预算的生成任务部署
扩散缓存是一种轻量级加速策略，通过复用去噪步骤间的中间特征来加速扩散变换器（DiTs），但现有方法存在根本性权衡：在线启发式规则缺乏全局预算感知，而静态调度缺乏实例自适应能力，无法灵活响应不同的运行时预算约束。为此，我们提出BAG（预算感知门控），一种将全局预算控制与动态实例自适应特征重用统一的新缓存策略。BAG不依赖人工规则，而是通过一个轻量门控网络，联合依据预算状态和局部轨迹反馈，动态决策每步是否执行完整计算或复用缓存特征。该策略通过离线到在线调度蒸馏训练，将离线搜索得到的调度决策迁移到紧凑的在线门控中。在FLUX.1-dev、Wan2.1和Qwen-Image-2512上的大量实验表明，BAG在不同加速等级下均持续优于现有先进缓存方法，且对不同分辨率、随机种子和引导尺度保持鲁棒性。代码将公开。
原文摘要 · Abstract (English)
Diffusion caching is a lightweight strategy that accelerates Diffusion Transformers (DiTs) by reusing intermediate features across denoising steps, but existing paradigms face a fundamental trade-off: online heuristics lack global budget awareness, whereas static schedules lack instance adaptivity and fail to flexibly adapt to varying runtime budget constraints. To bridge this gap, we present BAG (Budget-Aware Gating), a novel caching policy that unifies global budget pacing with dynamic, instance-adaptive feature reuse. Rather than relying on hand-crafted rules, BAG employs a lightweight gating network that dynamically decides whether to execute a full computation or reuse cached features at each step by jointly conditioning on the budget state and local trajectory feedback. We train this policy via offline-to-online schedule distillation, transferring the decision-making of offline-searched schedules into a compact online gate. Extensive experiments on FLUX.1-dev, Wan2.1, and Qwen-Image-2512 demonstrate that BAG consistently outperforms state-of-the-art caching methods across various speedup tiers while remaining robust across different resolutions, seeds, and guidance scales. Code will be released.
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。