Toxic Comment Classification Problem was a Kaggle Competition of 2018 link : https://www.kaggle.com/c/jigsaw-toxic-comment-classification-challenge#description
Data can be Downloaded from the above link
The code is a mixture of ANN and TidyText Models, so as to get a flavor of both the 1-D CNN(which sees pattern in the text) and word frequencies .
The gloVe6B word Embedding are used with 100 dimensions.
The final probability output is a weighted average of the above outputs