Diffusion Model Patching via Mixture-of-Prompts

• Main Authors: Ham, Seokil, Woo, Sangmin, Kim, Jin-Young, Go, Hyojun, Park, Byeongjun, Kim, Changick
• Format: Journal Article
• Language: English
• Published: 28.05.2024
• Subjects: Computer Science - Artificial Intelligence; Computer Science - Computer Vision and Pattern Recognition

We present Diffusion Model Patching (DMP), a simple method to boost the performance of pre-trained diffusion models that have already reached convergence, with a negligible increase in parameters. DMP inserts a small, learnable set of prompts into the model's input space while keeping the origi...