Twin-Merging: Dynamic Integration of Modular Expertise in Model Merging

TL;DR

Twin-Merging introduces a dynamic, modular approach to model merging that separates shared and exclusive knowledge, significantly improving performance and adaptability across diverse language and vision tasks.

Contribution

The paper proposes Twin-Merging, a novel method that modularizes and dynamically merges shared and exclusive knowledge, reducing interference and enhancing efficiency in model merging.

Findings

Achieves 28.34% average improvement in discriminative tasks.

Surpasses fine-tuned upper bounds on generative tasks.

Effectively handles heterogeneous data in model merging.

Abstract

In the era of large language models, model merging is a promising way to combine multiple task-specific models into a single multitask model without extra training. However, two challenges remain: (a) interference between different models and (b) heterogeneous data during testing. Traditional model merging methods often show significant performance gaps compared to fine-tuned models due to these issues. Additionally, a one-size-fits-all model lacks flexibility for diverse test data, leading to performance degradation. We show that both shared and exclusive task-specific knowledge are crucial for merging performance, but directly merging exclusive knowledge hinders overall performance. In view of this, we propose Twin-Merging, a method that encompasses two principal stages: (1) modularizing knowledge into shared and exclusive components, with compression to reduce redundancy and enhance efficiency; (2) dynamically merging shared and task-specific knowledge based on the input. This approach narrows the performance gap between merged and fine-tuned models and improves adaptability to heterogeneous data. Extensive experiments on $20$ datasets for both language and vision tasks demonstrate the effectiveness of our method, showing an average improvement of $28.34\%$ in absolute normalized score for discriminative tasks and even surpassing the fine-tuned upper bound on the generative tasks. Our implementation is available in \url{https://github.com/LZY-the-boys/Twin-Merging}