profane_subspaces Code for the WOAH 2021 submission "Modeling Profanity and Hate Speech in Social Media with Semantic Subspaces" The Bert model that we fine-tuned on multilingual Twitter data and used for the multilingual experiments can be found here.