ResearchTasksTargeted RefusalFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedMistral-7B Generation Evaluation SetVanilla97.01CA15Apr 15, 2026Llama2-7B Generation Evaluation SetVanilla93.14Completion Accuracy (CA)15Apr 15, 2026Targeted Refusal LLaMA2-13B-Chat (test)No Defense100BadNets11Mar 17, 2026Targeted Refusal LLaMA2-7B-Chat (test)FT100BadNets11Mar 17, 2026