SemRoDe: Macro Adversarial Training to Learn Representations That are Robust to Word-Level Attacks

Open in new window