bert-base-uncased, trained on fake news dataset. Input title, text, split with ////////////

Output: 1 - fake, 0 - real.