ScaleAI 's Collections
MHJ

MHJ

Dataset and RMU model weights for LLM Defenses Are Not Robust to Multi-Turn Human Jailbreaks Yet