用双曲空间控制文本生成，更稳定地过滤不安全内容
Not All Latent Spaces Are Flat: Hyperbolic Concept Control
- 用平行传输在双曲空间中调整文本嵌入，实现更精准控制
- 在4个安全基准上超越现有方法，适配4种主流生成模型
- 无需训练新模型，轻量适配即可提升生成安全性
随着现代文本到图像（T2I）模型日益逼近高度逼真的内容生成，不安全内容生成的风险随之上升，控制机制变得至关重要。现有方法通过在文本嵌入上施加欧几里得调整来引导模型避开不安全概念。本文提出双曲控制（HyCon）：一种基于平行传输的新控制机制，利用语义对齐的双曲表示空间，实现更具表现力和更稳定的概念操控。HyCon复用现成生成模型与先进的双曲文本编码器，通过轻量级适配器连接。在四个安全基准和四种T2I骨干模型上均达到当前最优性能，证明双曲引导是一种实用且灵活的可靠T2I生成方案。
原文摘要 · Abstract (English)
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.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。