Detecting Stealthy Backdoor Samples based on Intra-class Distance for Large Language Models

Open in new window