My shot at improving upon the ML literature results for the FNC dataset (2017). I take a two stage approach to the classification problem that beats the 2nd and 3rd place all time leaders. Paper describing my approach in detail and notebooks provided in this repo.