This project is based on Kaggle Competition: Toxic Comment Classification Challenge The challenge was to build a multi-headed model that’s capable of detecting different types of of toxicity like threats, obscenity, insults, and identity-based hate better than Perspective’s current models. Comments from Wikipedia’s talk page edits were used as dataset to train models. This was my first NLP Competition on Kaggle. Since everything was new to me, I learned a lot of new concepts, terms and techniques during the competiton.