ResearchTasksPrefix-risk rankingFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedTerminalBench (held-out)PrefixGuard (GRU)0.557AUPRC11May 8, 2026SkillsBench (held-out)PrefixGuard (GRU)53.3AUPRC11May 8, 2026tau^2-Bench (held-out)PrefixGuard (Transformer)71AUPRC11May 8, 2026WebArena (held-out)PrefixGuard (GRU)90AUPRC11May 8, 2026