Enhancing Out-of-Distribution Detection in Natural Language Understanding via Implicit Layer Ensemble

Open in new window