# ZJUKLAB at SemEval-2025 Task 4: Unlearning via Model Merging

Source: [ZJUKLAB at SemEval-2025 Task 4: Unlearning via Model Merging](<https://www.sota2.com/research/paper/68fa78241e209131cea76f80>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-10-26T04:32:25.977Z

Authors: Haoming Xu, Shuxun Wang, Yanqiu Zhao, Yi Zhong, Ziyan Jiang, Ningyuan Zhao, Shumin Deng, Huajun Chen, Ningyu Zhang

Published: 2025-03-27T02:03:25.000Z

## Abstract

This paper presents the ZJUKLAB team's submission for SemEval-2025 Task 4: Unlearning Sensitive Content from Large Language Models. This task aims to selectively erase sensitive knowledge from large language models, avoiding both over-forgetting and under-forgetting issues. We propose an unlearning system that leverages Model Merging (specifically TIES-Merging), combining two specialized models into a more balanced unlearned model. Our system achieves competitive results, ranking second among 26 teams, with an online score of 0.944 for Task Aggregate and 0.487 for overall Aggregate. In this paper, we also conduct local experiments and perform a comprehensive analysis of the unlearning process, examining performance trajectories, loss dynamics, and weight perspectives, along with several supplementary experiments, to understand the effectiveness of our method. Furthermore, we analyze the shortcomings of our method and evaluation metrics, emphasizing that MIA scores and ROUGE-based metrics alone are insufficient to fully evaluate successful unlearning. Finally, we emphasize the need for more comprehensive evaluation methodologies and rethinking of unlearning objectives in future research. Code is available at https://github.com/zjunlp/unlearn/tree/main/semeval25.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Machine Unlearning on MedQA QF=100](<https://www.sota2.com/research/sota/machine-unlearning-on-medqa-qf-100.md>): Machine Unlearning; MedQA QF=100; Forget Accuracy: 91.33.

- [Machine Unlearning on MedMCQA QF=1000](<https://www.sota2.com/research/sota/machine-unlearning-on-medmcqa-qf-1000.md>): Machine Unlearning; MedMCQA QF=1000; Forget Accuracy: 88.67.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2503.21088>)

[PDF](<https://arxiv.org/pdf/2503.21088.pdf>)