ResearchBenchmarksBinary Toxic Content Classification on Jigsaw-BLFollow99.27Attack Success RateNo AT58.720469.247779.77590.3023Apr 9, 2024Evaluation ResultsMethodMethodLinksAttack Success RateAvg QueriesAvg Perturbed WordsNo ATTraining=No AT, Search...Training=No AT, Search=unk, Time (s)=9392024.0499.2747.626.76AT1-unkTraining=AT1-unk, Sear...Training=AT1-unk, Search=unk, Time (s)=35462024.0460.28103.5413.03