Machine learningLLMs & Text

SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models

Hallucination Detection for LLMs: The paper presents SelfCheckGPT, a new approach for fact-checking black-box model responses without an external database, proving its superior ability to detect and rank factual and non-factual sentences.