Chain of Natural Language Inference for Reducing Large Language Model Ungrounded Hallucinations
This addresses the issue of ungrounded hallucinations in LLMs for industry applications, representing an incremental improvement with a plug-and-play framework.
The paper tackles the problem of large language models generating hallucinations not supported by provided sources by proposing a hierarchical framework using Chain of Natural Language Inference for detection and post-editing reduction, achieving state-of-the-art performance on hallucination detection and enhancing text quality without fine-tuning or domain-specific prompts.
Large language models (LLMs) can generate fluent natural language texts when given relevant documents as background context. This ability has attracted considerable interest in developing industry applications of LLMs. However, LLMs are prone to generate hallucinations that are not supported by the provided sources. In this paper, we propose a hierarchical framework to detect and mitigate such ungrounded hallucination. Our framework uses Chain of Natural Language Inference (CoNLI) for hallucination detection and hallucination reduction via post-editing. Our approach achieves state-of-the-art performance on hallucination detection and enhances text quality through rewrite, using LLMs without any fine-tuning or domain-specific prompt engineering. We show that this simple plug-and-play framework can serve as an effective choice for hallucination detection and reduction, achieving competitive performance across various contexts.