Gradient-Based Adversarial Training on Transformer Networks for Detecting Check-Worthy Factual Claims | Read Paper on Bytez