Same Question, Different Words: A Latent Adversarial Framework for Prompt Robustness

Publication
EMNLP 2025