Using a Wikipedia Dump we extracted the sentences containg the selected words. Then we passed these sentences through Elmo in order to generate contextualized vectors.