Couple solutions I made for the Google Jigsaw (toxicity severity rating competition)event held on Kaggle. My final ranking - 306/2302 teams. Preprocessing heavily involved the use of the GloVe dataset
JigsawCNN+RNN- A model utilising both Conv1D networks and GRUs to classify a given comment as toxic/non toxic.
JigsawSVM- A better (but way longer) solution to the jigsaw problem. Still utilised the GloVe (Global Vectorisation) dataset to vectorise the given sentence data