ResearchDatasetsHumanEval + AdvBenchHumanEval + AdvBenchFollowBenchmarksTaskDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyTaskDatasetSOTA resultTrendResultsMulti-agent code generation defense against sleeper agentsHumanEval + AdvBench (5:1 ratio)94.8Defense Success Rate (DSR)3
Multi-agent code generation defense against sleeper agentsHumanEval + AdvBench (5:1 ratio)94.8Defense Success Rate (DSR)3