We introduce a new loss function TripleEntropy, to improve classification performance for fine-tuning general knowledge pre-trained language models based on cross-entropy and SoftTriple loss. This loss function can improve the robust RoBERTa baseline model fine-tuned with cross-entropy loss by about (0.02% - 2.29%). Thorough tests on popular datasets indicate a steady gain. The fewer samples in the training dataset, the higher gain -- thus, for small-sized dataset it is 0.78%, for medium-sized -- 0.86% for large -- 0.20% and for extra-large 0.04%.
Cite
@article{arxiv.2112.08462,
title = {Applying SoftTriple Loss for Supervised Language Model Fine Tuning},
author = {Witold Sosnowski and Anna Wroblewska and Piotr Gawrysiak},
journal= {arXiv preprint arXiv:2112.08462},
year = {2022}
}