SG-Bench: Evaluating LLM Safety Generalization Across Diverse Tasks and Prompt Types

ELM
Re-assign community