4 papers
Distilling Safe LLM Systems via Soft Prompts for On Device Settings
Motasem Alfarra, Cristina Pinneri, Dana Kianfar +2
Deploying safe large language models (LLMs) on resource-constrained edge devices presents a critical challenge: while dual-model systems combining LLMs with guard models provide ef…
Masks Can Be Distracting: On Context Comprehension in Diffusion Language Models
Julianna Piskorz, Cristina Pinneri, Alvaro Correia +3
Masked Diffusion Language Models (MDLMs) have recently emerged as a promising alternative to Autoregressive Language Models (ARLMs), leveraging a denoising objective that, in princ…
Analyzing and Improving Chain-of-Thought Monitorability Through Information Theory
Usman Anwar, Tim Bakker, Dana Kianfar +2
Chain-of-thought (CoT) monitors are LLM-based systems that analyze reasoning traces to detect when outputs may exhibit attributes of interest, such as test-hacking behavior during…
Guarding the Meaning: Self-Supervised Training for Semantic Robustness in Guard Models
Cristina Pinneri, Christos Louizos
Guard models are a critical component of LLM safety, but their sensitivity to superficial linguistic variations remains a key vulnerability. We show that even meaning-preserving pa…