Due to the huge amount of parameters, fine-tuning of pretrained language models (PLMs) is prone to overfitting in the low resource scenarios. In this work, we present a novel method that operates on the hidden representations of a PLM to reduce overfitting. During fine-tuning, our method inserts random autoencoders between the hidden layers of a PLM, which transform activations from the previous layers into multi-view compressed representations before feeding them into the upper layers. The autoencoders are plugged out after fine-tuning, so our method does not add extra parameters or increase computation cost during inference. Our method demonstrates promising performance improvement across a wide range of sequence- and token-level low-resource NLP tasks.
@article{arxiv.2211.08794,
title = {Towards Robust Low-Resource Fine-Tuning with Multi-View Compressed Representations},
author = {Linlin Liu and Xingxuan Li and Megh Thakkar and Xin Li and Shafiq Joty and Luo Si and Lidong Bing},
journal= {arXiv preprint arXiv:2211.08794},
year = {2023}
}