HaloScope: Harnessing Unlabeled LLM Generations for Hallucination Detection
This addresses the critical issue of maintaining trust in LLM-generated content by providing a practical, annotation-free method for hallucination detection, though it is incremental as it builds on existing classification approaches.
The paper tackles the problem of detecting hallucinations in large language model (LLM) outputs by introducing HaloScope, a framework that uses unlabeled LLM generations to train a truthfulness classifier, achieving superior performance and outperforming competitors significantly.
The surge in applications of large language models (LLMs) has prompted concerns about the generation of misleading or fabricated information, known as hallucinations. Therefore, detecting hallucinations has become critical to maintaining trust in LLM-generated content. A primary challenge in learning a truthfulness classifier is the lack of a large amount of labeled truthful and hallucinated data. To address the challenge, we introduce HaloScope, a novel learning framework that leverages the unlabeled LLM generations in the wild for hallucination detection. Such unlabeled data arises freely upon deploying LLMs in the open world, and consists of both truthful and hallucinated information. To harness the unlabeled data, we present an automated membership estimation score for distinguishing between truthful and untruthful generations within unlabeled mixture data, thereby enabling the training of a binary truthfulness classifier on top. Importantly, our framework does not require extra data collection and human annotations, offering strong flexibility and practicality for real-world applications. Extensive experiments show that HaloScope can achieve superior hallucination detection performance, outperforming the competitive rivals by a significant margin. Code is available at https://github.com/deeplearningwisc/haloscope.