6/ ShieldGemma – offers a comprehensive suite of LLM-based safety content moderation models built on Gemma 2; includes classifiers for key harm types such as dangerous content, toxicity, hate speech, and more.

→ View original post on X — @dair_ai