Is What You Ask For What You Get? Investigating Concept Associations in Text-to-Image Models

TL;DR

This paper introduces Concept2Concept, a framework for auditing text-to-image models by analyzing concept associations in generated images, and provides an open-source visualization tool for easier interpretation.

Contribution

We propose a novel framework that characterizes model outputs using interpretable concepts, enabling systematic auditing of text-to-image models and prompt datasets.

Findings

Effective characterization of conditional distributions in T2I models

Demonstrated case studies with real-world prompt distributions

Open-source visualization tool for non-technical users

Abstract

Text-to-image (T2I) models are increasingly used in impactful real-life applications. As such, there is a growing need to audit these models to ensure that they generate desirable, task-appropriate images. However, systematically inspecting the associations between prompts and generated content in a human-understandable way remains challenging. To address this, we propose Concept2Concept, a framework where we characterize conditional distributions of vision language models using interpretable concepts and metrics that can be defined in terms of these concepts. This characterization allows us to use our framework to audit models and prompt-datasets. To demonstrate, we investigate several case studies of conditional distributions of prompts, such as user-defined distributions or empirical, real-world distributions. Lastly, we implement Concept2Concept as an open-source interactive visualization tool to facilitate use by non-technical end-users. A demo is available at https://tinyurl.com/Concept2ConceptDemo.