Not All Latent Spaces Are Flat: Hyperbolic Concept Control

TL;DR

HyCon introduces a hyperbolic control method for text-to-image models, enabling more expressive and stable concept manipulation to improve safety and reliability.

Contribution

It presents a novel hyperbolic control mechanism using parallel transport, enhancing concept steering in T2I models beyond Euclidean adjustments.

Findings

HyCon achieves state-of-the-art results on safety benchmarks.

It works across four T2I backbones.

Hyperbolic steering improves reliability of content generation.

Abstract

As modern text-to-image (T2I) models draw closer to synthesizing highly realistic content, the threat of unsafe content generation grows, and it becomes paramount to exercise control. Existing approaches steer these models by applying Euclidean adjustments to text embeddings, redirecting the generation away from unsafe concepts. In this work, we introduce hyperbolic control (HyCon): a novel control mechanism based on parallel transport that leverages semantically aligned hyperbolic representation space to yield more expressive and stable manipulation of concepts. HyCon reuses off-the-shelf generative models and a state-of-the-art hyperbolic text encoder, linked via a lightweight adapter. HyCon achieves state-of-the-art results across four safety benchmarks and four T2I backbones, showing that hyperbolic steering is a practical and flexible approach for more reliable T2I generation.