SafeRoute: Adaptive Model Selection for Efficient and Accurate Safety Guardrails in Large Language Models
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866908373873590272 |
|---|---|
| author | Lee, Seanie Lee, Dong Bok Wagner, Dominik Kang, Minki Seong, Haebin Bocklet, Tobias Lee, Juho Hwang, Sung Ju |
| author_facet | Lee, Seanie Lee, Dong Bok Wagner, Dominik Kang, Minki Seong, Haebin Bocklet, Tobias Lee, Juho Hwang, Sung Ju |
| contents | Deploying large language models (LLMs) in real-world applications requires robust safety guard models to detect and block harmful user prompts. While large safety guard models achieve strong performance, their computational cost is substantial. To mitigate this, smaller distilled models are used, but they often underperform on "hard" examples where the larger model provides accurate predictions. We observe that many inputs can be reliably handled by the smaller model, while only a small fraction require the larger model's capacity. Motivated by this, we propose SafeRoute, a binary router that distinguishes hard examples from easy ones. Our method selectively applies the larger safety guard model to the data that the router considers hard, improving efficiency while maintaining accuracy compared to solely using the larger safety guard model. Experimental results on multiple benchmark datasets demonstrate that our adaptive model selection significantly enhances the trade-off between computational cost and safety performance, outperforming relevant baselines. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_12464 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | SafeRoute: Adaptive Model Selection for Efficient and Accurate Safety Guardrails in Large Language Models Lee, Seanie Lee, Dong Bok Wagner, Dominik Kang, Minki Seong, Haebin Bocklet, Tobias Lee, Juho Hwang, Sung Ju Computation and Language Deploying large language models (LLMs) in real-world applications requires robust safety guard models to detect and block harmful user prompts. While large safety guard models achieve strong performance, their computational cost is substantial. To mitigate this, smaller distilled models are used, but they often underperform on "hard" examples where the larger model provides accurate predictions. We observe that many inputs can be reliably handled by the smaller model, while only a small fraction require the larger model's capacity. Motivated by this, we propose SafeRoute, a binary router that distinguishes hard examples from easy ones. Our method selectively applies the larger safety guard model to the data that the router considers hard, improving efficiency while maintaining accuracy compared to solely using the larger safety guard model. Experimental results on multiple benchmark datasets demonstrate that our adaptive model selection significantly enhances the trade-off between computational cost and safety performance, outperforming relevant baselines. |
| title | SafeRoute: Adaptive Model Selection for Efficient and Accurate Safety Guardrails in Large Language Models |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2502.12464 |