In this project, a model has been trained to recognize the positive or negative feeling related to each comment in the IMDB data. The movie review dataset, known as the IMDB dataset, contains 25,000 positive and negative reviews. This dataset can be accessed via this link: http://ai.stanford.edu/~amaas/data/sentiment To train this network, LSTM is used and a model is built whose input is a review and whose output is an agreement or disagreement. Unidirectional and bidirectional LSTM network with batch size = 50, a hidden layer with 64 memory units, a sigmoid activation function and learning rate = 0.01 was created separatly.