Does Fine-Tuning LLMs on New Knowledge Encourage Hallucinations?

Zorik Gekhman, Gal Yona, Roee Aharoni, Matan Eyal, Amir Feder, Roi Reichart, Jonathan Herzig


Abstract
When large language models are aligned via supervised fine-tuning, they may encounter new factual information that was not acquired through pre-training. It is often conjectured that this can teach the model the behavior of hallucinating factually incorrect responses, as the model is trained to generate facts that are not grounded in its pre-existing knowledge. In this work, we study the impact of such exposure to new knowledge on the capability of the fine-tuned model to utilize its pre-existing knowledge. To this end, we design a controlled setup, focused on closed-book QA, where we vary the proportion of the fine-tuning examples that introduce new knowledge. We demonstrate that large language models struggle to acquire new factual knowledge through fine-tuning, as fine-tuning examples that introduce new knowledge are learned significantly slower than those consistent with the model’s knowledge. However, we also find that as the examples with new knowledge are eventually learned, they linearly increase the model’s tendency to hallucinate. Taken together, our results highlight the risk in introducing new factual knowledge through fine-tuning, and support the view that large language models mostly acquire factual knowledge through pre-training, whereas fine-tuning teaches them to use it more efficiently.
Anthology ID:
2024.emnlp-main.444
Volume:
Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing
Month:
November
Year:
2024
Address:
Miami, Florida, USA
Editors:
Yaser Al-Onaizan, Mohit Bansal, Yun-Nung Chen
Venue:
EMNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
7765–7784
Language:
URL:
https://aclanthology.org/2024.emnlp-main.444
DOI:
Bibkey:
Cite (ACL):
Zorik Gekhman, Gal Yona, Roee Aharoni, Matan Eyal, Amir Feder, Roi Reichart, and Jonathan Herzig. 2024. Does Fine-Tuning LLMs on New Knowledge Encourage Hallucinations?. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 7765–7784, Miami, Florida, USA. Association for Computational Linguistics.
Cite (Informal):
Does Fine-Tuning LLMs on New Knowledge Encourage Hallucinations? (Gekhman et al., EMNLP 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.emnlp-main.444.pdf