IRIT-PREVISION AT HASOC 2020 : Fine-tuning BERT for Hate Speech and Offensive Content Identification

This paper describes the participation of the IRIT-PREVISION team at HASOC (Hate Speech and Offensive Content Identification in Indo-European Languages) 2020 shared task. Our approach is based on fine-tuning a pre-trained transformer based language model BERT (Bidirectional Encoder Representation from Transformer) [1]. We participated to the English sub-task A. We obtained a macro average F1 of 0.497 (self-computed).