arxivMay 11
GLiGuard: Schema-Conditioned Classification for LLM Safeguard
arXiv:2605.07982v1 Announce Type: new Abstract: Ensuring safe, policy-compliant outputs from large language models requires real-time content moderation that can scale across multiple safety dimensions. However, state-of-the-art guardrail models rely on autoregressive decoders with 7B--27B parameter