@article{thatisasuspiciousreactioninterpreting1, title = {"That Is a Suspicious Reaction!": Interpreting Logits Variation to Detect NLP Adversarial Attacks}, author = {Edoardo Mosca and Shreyash Agarwal and Javier Rando and Georg Groh}, year = {2022}, eprint = {2204.04636}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2204.04636v2}, }