Enhancing Self-Consistency and Performance of Pre-Trained Language Models through Natural Language Inference

Open in new window