Is Modularity Transferable? A Case Study through the Lens of Knowledge Distillation
This addresses a limitation in modular deep learning by enabling module reuse across models, which could reduce training costs and improve efficiency for NLP practitioners, though it appears incremental as it builds on existing PEFT and knowledge distillation methods.
The paper tackles the problem of transferring parameter-efficient fine-tuning (PEFT) modules between pre-trained language models, showing that a knowledge distillation-based approach enables transfer between same-family and incompatible models without increasing inference complexity, with experiments on tasks like Named Entity Recognition demonstrating initial potential.
The rise of Modular Deep Learning showcases its potential in various Natural Language Processing applications. Parameter-efficient fine-tuning (PEFT) modularity has been shown to work for various use cases, from domain adaptation to multilingual setups. However, all this work covers the case where the modular components are trained and deployed within one single Pre-trained Language Model (PLM). This model-specific setup is a substantial limitation on the very modularity that modular architectures are trying to achieve. We ask whether current modular approaches are transferable between models and whether we can transfer the modules from more robust and larger PLMs to smaller ones. In this work, we aim to fill this gap via a lens of Knowledge Distillation, commonly used for model compression, and present an extremely straightforward approach to transferring pre-trained, task-specific PEFT modules between same-family PLMs. Moreover, we propose a method that allows the transfer of modules between incompatible PLMs without any change in the inference complexity. The experiments on Named Entity Recognition, Natural Language Inference, and Paraphrase Identification tasks over multiple languages and PEFT methods showcase the initial potential of transferable modularity.