Chain-of-Scrutiny: Detecting Backdoor Attacks for Large Language Models

Open in new window