<ul><li>Recent research in hallucination detection in large language models (LLMs) has shown that LLMs' internal representations contain truthfulness hints that can be used for detector training.</li><li>However, the performance of these detectors is heavily dependent on predetermined tokens and fluctuates when working on free-form generations with varying lengths and sparse distributions of hallucinated entities.</li><li>To address this, a novel approach called HaMI is proposed, which enables robust detection of hallucinations through adaptive selection and learning of critical tokens that are most indicative of hallucinations.</li><li>Experimental results on four hallucination benchmarks demonstrate that HaMI outperforms existing state-of-the-art approaches.</li></ul>

Robust Hallucination Detection in LLMs via Adaptive Token Selection

Discover more