English

Learning Which Features Matter: RoBERTa Acquires a Preference for Linguistic Generalizations (Eventually)

Computation and Language 2020-10-13 v1

Abstract

One reason pretraining on self-supervised linguistic tasks is effective is that it teaches models features that are helpful for language understanding. However, we want pretrained models to learn not only to represent linguistic features, but also to use those features preferentially during fine-turning. With this goal in mind, we introduce a new English-language diagnostic set called MSGS (the Mixed Signals Generalization Set), which consists of 20 ambiguous binary classification tasks that we use to test whether a pretrained model prefers linguistic or surface generalizations during fine-tuning. We pretrain RoBERTa models from scratch on quantities of data ranging from 1M to 1B words and compare their performance on MSGS to the publicly available RoBERTa-base. We find that models can learn to represent linguistic features with little pretraining data, but require far more data to learn to prefer linguistic generalizations over surface ones. Eventually, with about 30B words of pretraining data, RoBERTa-base does demonstrate a linguistic bias with some regularity. We conclude that while self-supervised pretraining is an effective way to learn helpful inductive biases, there is likely room to improve the rate at which models learn which features matter.

Keywords

Cite

@article{arxiv.2010.05358,
  title  = {Learning Which Features Matter: RoBERTa Acquires a Preference for Linguistic Generalizations (Eventually)},
  author = {Alex Warstadt and Yian Zhang and Haau-Sing Li and Haokun Liu and Samuel R. Bowman},
  journal= {arXiv preprint arXiv:2010.05358},
  year   = {2020}
}

Comments

accepted at EMNLP 2020

R2 v1 2026-06-23T19:15:29.963Z