Further pre-training BERT and RoBERTa on next-sentence polarity prediction and a polarity-reversing variant of next sentence prediction improves negation reasoning by 1.8 to 9.3 accuracy points on CondaQA.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Making Language Models Robust Against Negation
Further pre-training BERT and RoBERTa on next-sentence polarity prediction and a polarity-reversing variant of next sentence prediction improves negation reasoning by 1.8 to 9.3 accuracy points on CondaQA.