用轻量编码器降低扩散模型轨迹弯曲，加速采样并提升画质。
Learning Quantized Adaptive Conditions for Diffusion Models
- 通过自适应条件与量化编码器减少微分方程轨迹弯曲。
- 仅6次函数评估即达CIFAR-10上5.14的FID，性能优异。
- 适合追求高效高质生成的图像生成研究者使用。
扩散模型中微分方程轨迹的曲率限制了其在少量函数评估（NFE）下的高质量图像生成能力。本文提出一种新颖有效的方法，通过自适应条件降低轨迹曲率。采用极轻量的量化编码器，训练参数仅增加1%，无需额外正则化项，却显著提升样本质量。该方法加速了ODE采样过程，同时保持SDE技术在下游图像编辑任务中的能力。大量实验验证，本方法可在极低采样成本下生成高质量结果：仅需6次NFE，CIFAR-10上实现5.14 FID，FFHQ 64x64上为6.91 FID，AFHQv2上为3.10 FID。
原文摘要 · Abstract (English)
The curvature of ODE trajectories in diffusion models hinders their ability to generate high-quality images in a few number of function evaluations (NFE). In this paper, we propose a novel and effective approach to reduce trajectory curvature by utilizing adaptive conditions. By employing a extremely light-weight quantized encoder, our method incurs only an additional 1% of training parameters, eliminates the need for extra regularization terms, yet achieves significantly better sample quality. Our approach accelerates ODE sampling while preserving the downstream task image editing capabilities of SDE techniques. Extensive experiments verify that our method can generate high quality results under extremely limited sampling costs. With only 6 NFE, we achieve 5.14 FID on CIFAR-10, 6.91 FID on FFHQ 64x64 and 3.10 FID on AFHQv2.
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。