解决扩散模型蒸馏后图像质量下降问题，无需训练即可提升多样性。
EMPURPLE: A Free Lunch for Diffusion Distillation based on the Information Bottleneck
- 通过信息瓶颈分析发现早期步骤速度场重定向导致分布失配。
- 提出EMPURPLE方法，复用原模型中间潜变量，提升生成多样性。
- 适用于各类蒸馏模型，不依赖训练，显著改善FID得分。
扩散模型虽生成质量优异，但推理成本高昂。扩散蒸馏可减少采样步骤，但包括SDXL-Lightning和分布匹配蒸馏在内的许多方法存在弗雷切特起始距离（FID）下降问题。我们通过类似PAC的泛化界分析指出：激进的早期步骤速度场重定向使蒸馏目标更难学习，扩大了训练与测试间的差距，导致中间噪声潜变量在训练与推理时分布不一致。我们通过实验证明中间特征和最终输出多样性降低。为此，提出EMPURPLE——一种无需训练的通用方法，复用原始模型采样的中间潜变量。该方法在DMD2、Hyper-SD、FlashSD和SDXL-Lightning上均提升FID 7%至20%。项目地址：https://github.com/TheLovesOfLadyPurple/Empurple-Training-Free-Algorithm-To-enhance-Diversity-of-The-Diffusion-Distillation-Model
原文摘要 · Abstract (English)
Diffusion models achieve impressive image-generation quality but remain expensive at inference time. Diffusion distillation reduces sampling steps, yet many distilled models, including SDXL-Lightning and distribution matching distillation methods, suffer from degraded Fréchet Inception Distance (FID). We analyze this phenomenon through a PAC-style generalization bound. Our analysis suggests that aggressive early-step redirection of the velocity field makes the distillation target harder to learn, enlarging the train-test gap. As a result, early-step output distributions differ between training and inference, causing distribution mismatch in the intermediate noisy latent used as next-step inputs. We empirically validate this mechanism by showing reduced diversity in both intermediate features and final outputs. To address this issue, we propose EMPURPLE, a simple training-free method that recycles intermediate latents sampled from the original model. EMPURPLE is model-agnostic and improves FID by 7\% to 20\% across DMD2, Hyper-SD, FlashSD, and SDXL-Lightning. The repo is: https://github.com/TheLovesOfLadyPurple/Empurple-Training-Free-Algorithm-To-enhance-Diversity-of-The-Diffusion-Distillation-Model
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。