DDIL：基于模仿学习的多样性增强扩散蒸馏
DDIL: Diversity Enhancing Diffusion Distillation With Imitation Learning
AI总结：
本文提出DDIL框架，通过模仿学习结合数据分布与学生分布训练，解决扩散蒸馏中的协变量偏移，提升生成多样性与稳定性，优于PD、LCM和DMD2等基线方法。
摘要：
扩散模型在生成建模（如文本到图像）方面表现出色，但采样需要多次去噪网络传递，限制了其实用性。渐进式蒸馏或一致性蒸馏等努力通过减少传递次数显示出潜力，但以牺牲生成样本的质量为代价。在这项工作中，我们识别出协变量偏移是导致多步蒸馏模型在推理时因复合误差而性能不佳的原因之一。为了解决协变量偏移，我们在模仿学习（DDIL）框架内构建扩散蒸馏，并增强蒸馏扩散模型的训练分布，既涵盖数据分布（前向扩散）又涵盖学生诱导分布（后向扩散）。在数据分布上训练有助于通过保留边际数据分布来多样化生成，而在学生分布上训练则通过纠正协变量偏移来解决复合误差。此外，我们采用反射扩散公式进行蒸馏，并展示了在不同蒸馏方法中改进的性能和稳定的训练。我们表明，DDIL一致性在渐进式蒸馏（PD）、潜在一致性模型（LCM）和分布匹配蒸馏（DMD2）的基线算法上有所改进。
英文摘要：
Diffusion models excel at generative modeling (e.g., text-to-image) but sampling requires multiple denoising network passes, limiting practicality. Efforts such as progressive distillation or consistency distillation have shown promise by reducing the number of passes at the expense of quality of the generated samples. In this work we identify co-variate shift as one of reason for poor performance of multi-step distilled models from compounding error at inference time. To address co-variate shift, we formulate diffusion distillation within imitation learning (DDIL) framework and enhance training distribution for distilling diffusion models on both data distribution (forward diffusion) and student induced distributions (backward diffusion). Training on data distribution helps to diversify the generations by preserving marginal data distribution and training on student distribution addresses compounding error by correcting covariate shift. In addition, we adopt reflected diffusion formulation for distillation and demonstrate improved performance, stable training across different distillation methods. We show that DDIL consistency improves on baseline algorithms of progressive distillation (PD), Latent consistency models (LCM) and Distribution Matching Distillation (DMD2).
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
