Raise a Child in Large Language Model: Towards Effective and Generalizable Fine-tuning
Abstract
Recent pretrained language models extend from millions to billions of parameters. Thus the need to fine-tune an extremely large pretrained model with a limited training corpus arises in various downstream tasks. In this paper, we propose a straightforward yet effective fine-tuning technique, Child-Tuning, which updates a subset of parameters (called child network) of large pretrained models via strategically masking out the gradients of the non-child network during the backward process. Experiments on various downstream tasks in GLUE benchmark show that Child-Tuning consistently outperforms the vanilla fine-tuning by 1.5~8.6 average score among four different pretrained models, and surpasses the prior fine-tuning techniques by 0.6~1.3 points. Furthermore, empirical results on domain transfer and task transfer show that Child-Tuning can obtain better generalization performance by large margins.
Keywords
Cite
@article{arxiv.2109.05687,
title = {Raise a Child in Large Language Model: Towards Effective and Generalizable Fine-tuning},
author = {Runxin Xu and Fuli Luo and Zhiyuan Zhang and Chuanqi Tan and Baobao Chang and Songfang Huang and Fei Huang},
journal= {arXiv preprint arXiv:2109.05687},
year = {2021}
}
Comments
Accepted as a long paper to EMNLP 2021 Main Conference