The objective of this assignment is to build a bag of n-grams model for predicting the sentiment of the movie reviewers given the textual review for the movie. The IMDB Movie review dataset consisting of 25000 train and 25000 test movie reviews which can be classified as positive or negative. The train dataset is split into 20000 train examples and 5000 validation examples. In this assignment, we will do ablation study on different tokenization schemes, model hyper-parameters and optimization hyper-parameters.