ResearchTasksEmbedded Direct AttackEmbedded Direct AttackFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedSamsum ⊕ HEx-PHILisa52.4Utility12Jun 30, 2026GSM8K ⊕ BeaverTailsSFT74.7Utility12Jun 30, 2026GSM8K ⊕ HEx-PHISFT76.1Utility12Jun 30, 2026