An Interpretable N-gram Perplexity Threat Model for Large Language Model Jailbreaks

Devs

An Interpretable N-gram Perplexity Threat Model for Large Language Model Jailbreaks | Read Paper on Bytez