Avoiding Generative Model Writer's Block With Embedding Nudging

TL;DR

This paper introduces a technique for steering generative image models away from unwanted concepts, such as memorized images, without compromising output quality, thereby enhancing control and usability.

Contribution

The paper presents a novel embedding nudging method that prevents specific image generations in latent diffusion models while preserving overall image quality.

Findings

Successfully prevents memorized image generation

Maintains comparable image quality and relevance

Effective in mitigating privacy and safety concerns

Abstract

Generative image models, since introduction, have become a global phenomenon. From new arts becoming possible to new vectors of abuse, many new capabilities have become available. One of the challenging issues with generative models is controlling the generation process specially to prevent specific generations classes or instances . There are several reasons why one may want to control the output of generative models, ranging from privacy and safety concerns to application limitations or user preferences To address memorization and privacy challenges, there has been considerable research dedicated to filtering prompts or filtering the outputs of these models. What all these solutions have in common is that at the end of the day they stop the model from producing anything, hence limiting the usability of the model. In this paper, we propose a method for addressing this usability issue by making it possible to steer away from unwanted concepts (when detected in model's output) and still generating outputs. In particular we focus on the latent diffusion image generative models and how one can prevent them to generate particular images while generating similar images with limited overhead. We focus on mitigating issues like image memorization, demonstrating our technique's effectiveness through qualitative and quantitative evaluations. Our method successfully prevents the generation of memorized training images while maintaining comparable image quality and relevance to the unmodified model.