杠杆(统计)
计算机科学
对抗制
忠诚
蒸馏
一致性(知识库)
图像(数学)
编码(集合论)
生成语法
扩散
比例(比率)
人工智能
采样(信号处理)
算法
机器学习
计算机视觉
程序设计语言
电信
化学
物理
有机化学
集合(抽象数据类型)
滤波器(信号处理)
量子力学
热力学
作者
Axel Sauer,Dominik Lorenz,Andreas Blattmann,Robin Rombach
出处
期刊:Cornell University - arXiv
日期:2023-01-01
标识
DOI:10.48550/arxiv.2311.17042
摘要
We introduce Adversarial Diffusion Distillation (ADD), a novel training approach that efficiently samples large-scale foundational image diffusion models in just 1-4 steps while maintaining high image quality. We use score distillation to leverage large-scale off-the-shelf image diffusion models as a teacher signal in combination with an adversarial loss to ensure high image fidelity even in the low-step regime of one or two sampling steps. Our analyses show that our model clearly outperforms existing few-step methods (GANs, Latent Consistency Models) in a single step and reaches the performance of state-of-the-art diffusion models (SDXL) in only four steps. ADD is the first method to unlock single-step, real-time image synthesis with foundation models. Code and weights available under https://github.com/Stability-AI/generative-models and https://huggingface.co/stabilityai/ .
科研通智能强力驱动
Strongly Powered by AbleSci AI