Enhancing Hallucination Detection through Perturbation-Based Synthetic Data Generation in System Responses

Dongxu Zhang, Varun Gangal, Barrett Lattimer, Yi Yang


Abstract
Detecting hallucinations in large language model (LLM) outputs is pivotal, yet traditional fine-tuning for this classification task is impeded by the expensive and quickly outdated annotation process, especially across numerous vertical domains and in the face of rapid LLM advancements. In this study, we introduce an approach that automatically generates both faithful and hallucinated outputs by rewriting system responses. Experimental findings demonstrate that a T5-base model, fine-tuned on our generated dataset, surpasses state-of-the-art zero-shot detectors and existing synthetic generation methods in both accuracy and latency, indicating efficacy of our approach.
Anthology ID:
2024.findings-acl.789
Volume:
Findings of the Association for Computational Linguistics: ACL 2024
Month:
August
Year:
2024
Address:
Bangkok, Thailand
Editors:
Lun-Wei Ku, Andre Martins, Vivek Srikumar
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
13321–13332
Language:
URL:
https://aclanthology.org/2024.findings-acl.789
DOI:
10.18653/v1/2024.findings-acl.789
Bibkey:
Cite (ACL):
Dongxu Zhang, Varun Gangal, Barrett Lattimer, and Yi Yang. 2024. Enhancing Hallucination Detection through Perturbation-Based Synthetic Data Generation in System Responses. In Findings of the Association for Computational Linguistics: ACL 2024, pages 13321–13332, Bangkok, Thailand. Association for Computational Linguistics.
Cite (Informal):
Enhancing Hallucination Detection through Perturbation-Based Synthetic Data Generation in System Responses (Zhang et al., Findings 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.findings-acl.789.pdf