Evaluating Evaluation Metrics -- The Mirage of Hallucination Detection
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866916998922895360 |
|---|---|
| author | Kulkarni, Atharva Zhang, Yuan Moniz, Joel Ruben Antony Ge, Xiou Tseng, Bo-Hsiang Piraviperumal, Dhivya Swayamdipta, Swabha Yu, Hong |
| author_facet | Kulkarni, Atharva Zhang, Yuan Moniz, Joel Ruben Antony Ge, Xiou Tseng, Bo-Hsiang Piraviperumal, Dhivya Swayamdipta, Swabha Yu, Hong |
| contents | Hallucinations pose a significant obstacle to the reliability and widespread adoption of language models, yet their accurate measurement remains a persistent challenge. While many task- and domain-specific metrics have been proposed to assess faithfulness and factuality concerns, the robustness and generalization of these metrics are still untested. In this paper, we conduct a large-scale empirical evaluation of 6 diverse sets of hallucination detection metrics across 4 datasets, 37 language models from 5 families, and 5 decoding methods. Our extensive investigation reveals concerning gaps in current hallucination evaluation: metrics often fail to align with human judgments, take an overtly myopic view of the problem, and show inconsistent gains with parameter scaling. Encouragingly, LLM-based evaluation, particularly with GPT-4, yields the best overall results, and mode-seeking decoding methods seem to reduce hallucinations, especially in knowledge-grounded settings. These findings underscore the need for more robust metrics to understand and quantify hallucinations, and better strategies to mitigate them. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_18114 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Evaluating Evaluation Metrics -- The Mirage of Hallucination Detection Kulkarni, Atharva Zhang, Yuan Moniz, Joel Ruben Antony Ge, Xiou Tseng, Bo-Hsiang Piraviperumal, Dhivya Swayamdipta, Swabha Yu, Hong Computation and Language Artificial Intelligence Machine Learning Hallucinations pose a significant obstacle to the reliability and widespread adoption of language models, yet their accurate measurement remains a persistent challenge. While many task- and domain-specific metrics have been proposed to assess faithfulness and factuality concerns, the robustness and generalization of these metrics are still untested. In this paper, we conduct a large-scale empirical evaluation of 6 diverse sets of hallucination detection metrics across 4 datasets, 37 language models from 5 families, and 5 decoding methods. Our extensive investigation reveals concerning gaps in current hallucination evaluation: metrics often fail to align with human judgments, take an overtly myopic view of the problem, and show inconsistent gains with parameter scaling. Encouragingly, LLM-based evaluation, particularly with GPT-4, yields the best overall results, and mode-seeking decoding methods seem to reduce hallucinations, especially in knowledge-grounded settings. These findings underscore the need for more robust metrics to understand and quantify hallucinations, and better strategies to mitigate them. |
| title | Evaluating Evaluation Metrics -- The Mirage of Hallucination Detection |
| topic | Computation and Language Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2504.18114 |