Backdoor Removal for Generative Large Language Models

arXiv:2405.07667v1 Announce Type: new
Abstract: With rapid advances, generative large language models (LLMs) dominate various Natural Language Processing (NLP) tasks from understanding to reasoning. Yet, language models' inherent vulnerabilities may be exacerbated due to increased accessibility and unrestricted model training on massive textual data from the Internet. A malicious adversary may publish poisoned data online and conduct backdoor attacks on the victim LLMs pre-trained on the poisoned data. Backdoored LLMs behave innocuously for normal queries and generate harmful responses when …

accessibility adversary arxiv backdoor cs.cl cs.cr data generative internet language language models large llms malicious may model training natural natural language natural language processing nlp rapid reasoning training understanding vulnerabilities