Multi-concept Model Immunization through Differentiable Model Merging

TL;DR

This paper introduces a novel multi-concept model immunization method using differentiable model merging to prevent fine-tuning on harmful applications, addressing a gap in existing single-concept approaches.

Contribution

It proposes a new immunization algorithm that learns a single difficult initialization for multiple concepts via differentiable weight merging.

Findings

Effective multi-concept immunization demonstrated in experiments.

Generalizes prior single-concept immunization to multiple concepts.

Improves robustness against harmful fine-tuning.

Abstract

Model immunization is an emerging direction that aims to mitigate the potential risk of misuse associated with open-sourced models and advancing adaptation methods. The idea is to make the released models' weights difficult to fine-tune on certain harmful applications, hence the name ``immunized''. Recent work on model immunization focuses on the single-concept setting. However, models need to be immunized against multiple concepts in real-world situations. To address this gap, we propose an immunization algorithm that, simultaneously, learns a single ``difficult initialization'' for adaptation methods over a set of concepts. We achieve this by incorporating a differentiable merging layer that combines a set of model weights adapted over multiple concepts. In our experiments, we demonstrate the effectiveness of multi-concept immunization by generalizing prior work's experiment setup of re-learning and personalization adaptation to multiple concepts.