发表机构
Tsinghua University（清华大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出QuAKE，一种量化感知卡尔曼估计器，利用轨迹历史在线估计全精度输出，以校正量化扩散采样误差，无需修改网络并支持高阶ODE采样器，显著减少分布差异。
AI 中文摘要
量化提供了一条以降低内存和计算开销来部署扩散模型的实用途径，但激进的压缩可能导致量化输出与全精度对应输出产生显著偏差。采样阶段的校正方法旨在采样过程中补偿此类偏差，然而现有方法主要依赖局部信息，未能充分利用轨迹历史，从而限制了其纠正跨时间步传播误差的能力。在本工作中，我们将使用量化去噪器的采样过程建模为一个在线估计问题，利用量化去噪器输出的历史来恢复采样器所需的全精度输出。我们提出QuAKE，一种量化感知卡尔曼估计器，它结合了平滑轨迹先验与条件高斯观测模型。在每个采样步骤中，QuAKE以闭式形式递归更新输出窗口上的后验分布，并将其后验均值馈送给采样器。QuAKE是一种轻量级即插即用的校正器，无需对量化网络进行任何修改，并自然支持任意高阶多步ODE采样器。在W4A4量化文本到图像扩散模型上的实验表明，QuAKE在减少与全精度采样的分布差异方面持续优于现有方法。
英文摘要
Quantization offers a practical path to deploying diffusion models with reduced memory and computation, but aggressive compression can cause quantized outputs to deviate substantially from their full-precision counterparts. Sampling-stage correction methods seek to compensate for such deviations during sampling, but existing approaches rely primarily on local information and underexploit trajectory history, limiting their ability to correct errors that propagate across timesteps. In this work, we formulate sampling with a quantized denoiser as an online estimation problem, using the history of quantized denoiser outputs to recover the underlying full-precision outputs required by the sampler. We propose QuAKE, a Quantization-Aware Kalman Estimator that combines a smooth trajectory prior with a conditional Gaussian observation model. At each sampling step, QuAKE recursively updates the posterior over the output window in closed form and feeds its posterior mean to the sampler. QuAKE is a lightweight plug-and-play corrector that requires no modification to the quantized network and naturally supports arbitrary high-order multistep ODE samplers. Experiments across W4A4-quantized text-to-image diffusion models show that QuAKE consistently outperforms existing methods in reducing the distributional discrepancy from full-precision sampling.
Comments20 pages, 6 figures, 2 tables
