Layerwise Anomaly This repository contains the source code and data for our ACL 2021 paper: "How is BERT surprised? Layerwise detection of linguistic