watched a movie? lets give a review and let model tell wheather its a positve or negative. the models were trained on Large Movie review dataset. right now a simple model NN which take use bag of words method. the achieved acuuracy is about 84%, which is not bad for such a simple architecture. you will also find in the script that "not boring" and "boring" both are classified as negative, and this is the limitations of these type of non sequential models.