1 paper · 1 filter
Benjamin Matthias Ruppik, Michael Heck, Carel van Niekerk +5
A common approach for sequence tagging tasks based on contextual word representations is to train a machine learning classifier directly on these embedding vectors. This approach h…