Semantic Energy: Detecting LLM Hallucination Beyond Entropy
This work addresses the critical issue of hallucination detection for reliable LLM deployment, representing an incremental advancement over existing methods like semantic entropy.
The paper tackles the problem of detecting hallucinations in Large Language Models by introducing Semantic Energy, a novel uncertainty estimation framework that operates on logits to better capture model confidence, resulting in significant improvements in hallucination detection across multiple benchmarks.
Large Language Models (LLMs) are being increasingly deployed in real-world applications, but they remain susceptible to hallucinations, which produce fluent yet incorrect responses and lead to erroneous decision-making. Uncertainty estimation is a feasible approach to detect such hallucinations. For example, semantic entropy estimates uncertainty by considering the semantic diversity across multiple sampled responses, thus identifying hallucinations. However, semantic entropy relies on post-softmax probabilities and fails to capture the model's inherent uncertainty, causing it to be ineffective in certain scenarios. To address this issue, we introduce Semantic Energy, a novel uncertainty estimation framework that leverages the inherent confidence of LLMs by operating directly on logits of penultimate layer. By combining semantic clustering with a Boltzmann-inspired energy distribution, our method better captures uncertainty in cases where semantic entropy fails. Experiments across multiple benchmarks show that Semantic Energy significantly improves hallucination detection and uncertainty estimation, offering more reliable signals for downstream applications such as hallucination detection.