Can I read Aegis Safety Data Set on EtoBox?
Aegis Safety Data Set by arakdamienvincent is a document available to read on EtoBox.
What is Aegis Safety Data Set about?
The document presents AEGIS, an online adaptive AI content safety moderation system utilizing a comprehensive risk taxonomy and a newly curated dataset of human-LLM interactions. It introduces AEGIS SAFETY EXPERTS, a suite of LLM-based models that outperform existing safety models while demonstrating robustness against various adversarial attacks. The framework employs a no-regret online adaptation approach to dynamically adjust content moderation based on context and evolving safety policies.
- Author
- arakdamienvincent
- Language
- EN