Making Language Models Robust Against Negation

Abstract: Negation has been a long-standing challenge for LLMs. Previous studies have shown that they struggle with negation in many natural language understanding tasks. In this work, we propose a self-supervised method to make LLMs more robust against negation. We introduce a novel task, Next Sentence Polarity Prediction (NSPP), and a variation of the Next Sentence Prediction (NSP) task. We show that BERT and RoBERTa further pre-trained on our tasks outperform the off-the-shelf versions on nine negation-related benchmarks. Most notably, our pre-training tasks yield between 1.8% and 9.1% improvement on CondaQA, a large question-answering corpus requiring reasoning over negation.