@inproceedings{ye-etal-2025-analyzing,
    title = "Analyzing the Effects of Supervised Fine-Tuning on Model Knowledge from Token and Parameter Levels",
    author = "Ye, Junjie  and
      Yang, Yuming  and
      Nan, Yang  and
      Li, Shuo  and
      Zhang, Qi  and
      Gui, Tao  and
      Huang, Xuanjing  and
      Wang, Peng  and
      Shi, Zhongchao  and
      Fan, Jianping",
    editor = "Christodoulopoulos, Christos  and
      Chakraborty, Tanmoy  and
      Rose, Carolyn  and
      Peng, Violet",
    booktitle = "Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing",
    month = nov,
    year = "2025",
    address = "Suzhou, China",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-emnlp/2025.emnlp-main.25/",
    pages = "471--513",
    ISBN = "979-8-89176-332-6",
    abstract = "Large language models (LLMs) acquire substantial world knowledge during pre-training, which is further shaped by post-training techniques such as supervised fine-tuning (SFT). However, the impact of SFT on a model{'}s knowledge remains underexplored, limiting our ability to control knowledge behavior in fine-tuned models. To address this gap, we evaluate closed-book question answering (CBQA) performance across five LLMs from the LLaMA-2 and LLaMA-3 families. Surprisingly, models fine-tuned on 1,920 samples perform up to 14{\%} worse than those fine-tuned on only 240 samples. Furthermore, varying the level of knowledge mastery in the fine-tuning data leads to performance fluctuations of over 12{\%}. To investigate these effects, we analyze model behavior at both the token and parameter levels. Our analysis reveals that up to 90{\%} of parameter updates during SFT do not contribute to knowledge enhancement. Restoring these updates can improve performance on the CBQA task, depending on the characteristics of the fine-tuning data. These insights offer practical guidance for developing fine-tuning strategies that more effectively strengthen model knowledge."
}