GradSafe: Detecting Unsafe Prompts for LLMs via Safety-Critical Gradient Analysis

Add code
Feb 21, 2024

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: