A Split-and-Privatize Framework for Large Language Model Fine-Tuning
Abstract: Fine-tuning is a prominent technique to adapt a pre-trained language model to downstream scenarios. In parameter-efficient fine-tuning, only a small subset of modules are trained over the downstream datasets, while leaving the rest of the pre-trained model frozen to save computation resources. In recent years, a popular productization form arises as Model-as-a-Service (MaaS), in which vendors provide abundant pre-trained language models, server resources and core functions, and customers can fine-tune, deploy and invoke their customized model by accessing the one-stop MaaS with their own private dataset. In this paper, we identify the model and data privacy leakage risks in MaaS fine-tuning, and propose a Split-and-Privatize (SAP) framework, which manage to mitigate the privacy issues by adapting the existing split learning architecture. Furthermore, we propose a contributing-token-identification (CTI) method to alleviate the utility degradation caused by privatization. The proposed framework is sufficiently investigated by experiments, and the results indicate that it can enhance the empirical privacy by $68\%$ at the cost of $1\%$ model performance degradation on the Stanford Sentiment Treebank dataset.
Paper Type: Long
Research Area: Language Modeling
Research Area Keywords: security and privacy, fine-tuning
Languages Studied: English
Submission Number: 1357
Loading