ResearchDatasetsTChatTChatFollowBenchmarksTaskDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyTaskDatasetSOTA resultTrendResultsHarmful prompt detectionTChat76.51F1 Score17