To Tune or Not To Tune? How About the Best of Both Worlds?
This addresses the challenge of efficient model adaptation for NLP researchers, but it is incremental as it builds on existing fine-tuning approaches.
The paper tackles the problem of adapting pre-trained language models like BERT to downstream tasks by proposing a new method that first trains with frozen parameters and then fine-tunes the entire model, achieving improvements such as 4.7% accuracy in semantic similarity.
The introduction of pre-trained language models has revolutionized natural language research communities. However, researchers still know relatively little regarding their theoretical and empirical properties. In this regard, Peters et al. perform several experiments which demonstrate that it is better to adapt BERT with a light-weight task-specific head, rather than building a complex one on top of the pre-trained language model, and freeze the parameters in the said language model. However, there is another option to adopt. In this paper, we propose a new adaptation method which we first train the task model with the BERT parameters frozen and then fine-tune the entire model together. Our experimental results show that our model adaptation method can achieve 4.7% accuracy improvement in semantic similarity task, 0.99% accuracy improvement in sequence labeling task and 0.72% accuracy improvement in the text classification task.