ResearchTasksSecret loyalty evaluationFor AI assistantsMarkdownContent indexSecret loyalty evaluationFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedWildChat final training-evaluation step (held-out)Narrow Secret Loyalty (7B Positive-Only ablation)77Activation Rate4May 11, 2026
WildChat final training-evaluation step (held-out)Narrow Secret Loyalty (7B Positive-Only ablation)77Activation Rate4May 11, 2026