Attention-based Contrastive Learning for Winograd Schemas
This addresses commonsense reasoning in NLP, with incremental improvements in unsupervised learning for specific tasks.
The paper tackled the Winograd Schema Challenge by extending contrastive learning to Transformer attention, resulting in a self-supervised framework that outperformed comparable unsupervised methods and sometimes surpassed supervised ones.
Self-supervised learning has recently attracted considerable attention in the NLP community for its ability to learn discriminative features using a contrastive objective. This paper investigates whether contrastive learning can be extended to Transfomer attention to tackling the Winograd Schema Challenge. To this end, we propose a novel self-supervised framework, leveraging a contrastive loss directly at the level of self-attention. Experimental analysis of our attention-based models on multiple datasets demonstrates superior commonsense reasoning capabilities. The proposed approach outperforms all comparable unsupervised approaches while occasionally surpassing supervised ones.