You can find the dataset on the kaggle :- https://www.kaggle.com/lakshmi25npathi/imdb-dataset-of-50k-movie-reviews
Train the file and then model will automatically saved with the best accuraccy. this model gives me accuracy upto 93%
And then for checking the results of the accuracy you can run the app.py file which will lead you to the flask server there you can check the positive reviews and negative reviews of the statements of the movies.
you can download the Bert_base_uncase/cased on hugging face itself.