Dataset from kaggle IMDB reviews (IMDB dataset) was used. It contains each of 5000 positive and negative reviews respectively.