Abstract
Despite improvements in performances on different natural language generation tasks, deep neural models are prone to hallucinating facts that are incorrect or nonexistent. Different hypotheses are proposed and examined separately for different tasks, but no systematic explanations are available across these tasks. In this study, we draw connections between hallucinations and predictive uncertainty in conditional language generation. We investigate their relationship in both image captioning and data-to-text generation and propose a simple extension to beam search to reduce hallucination. Our analysis shows that higher predictive uncertainty corresponds to a higher chance of hallucination. Epistemic uncertainty is more indicative of hallucination than aleatoric or total uncertainties. It helps to achieve better results of trading performance in standard metric for less hallucination with the proposed beam search variant.- Anthology ID:
- 2021.eacl-main.236
- Volume:
- Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume
- Month:
- April
- Year:
- 2021
- Address:
- Online
- Editors:
- Paola Merlo, Jorg Tiedemann, Reut Tsarfaty
- Venue:
- EACL
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 2734–2744
- Language:
- URL:
- https://aclanthology.org/2021.eacl-main.236
- DOI:
- 10.18653/v1/2021.eacl-main.236
- Cite (ACL):
- Yijun Xiao and William Yang Wang. 2021. On Hallucination and Predictive Uncertainty in Conditional Language Generation. In Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume, pages 2734–2744, Online. Association for Computational Linguistics.
- Cite (Informal):
- On Hallucination and Predictive Uncertainty in Conditional Language Generation (Xiao & Wang, EACL 2021)
- PDF:
- https://preview.aclanthology.org/nschneid-patch-4/2021.eacl-main.236.pdf
- Data
- ToTTo