Movie review 

Purpose: Predicting whether a movie review is positive or negative.

Dataset : IMDB. Include 50000 highly polarized reviews with 50% positive reviews and 50% negative reviews.

Development brief : Limit the amount of data loaded to exclude rare words. Use multi-hot coding to change the data type. Then vectorize the data.

Model :

To find where is the overfiting point and retraining the model in New epoch. And the result of it :

Predict :

Evaluate the accuracy :

Detail :

https://colab.research.google.com/drive/1hizHthjPBv8M7SXhJNxEsK_Ur3FeyQFu?usp=share_link