About this document
2025 Emnlp-Main 1761 by oldcity962 is a document available to read on EtoBox.
This document critiques the effectiveness of current hallucination detection methods in large language models (LLMs), particularly highlighting the inadequacies of the ROUGE metric, which misaligns with human judgments due to its low precision. Through human-aligned evaluations, the authors demonstrate that simpler heuristics based on response length can outperform complex detection techniques, revealing a significant overestimation of existing methods. The study advocates for the adoption of more robust, s
- Author
- oldcity962
- Language
- EN