ResearchDatasetsLlama3-8BLlama3-8BFollowBenchmarksTaskDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyTaskDatasetSOTA resultTrendResultsWatermark DetectionLlama3-8B generated text max 30 tokens0.008Detection Time (s)6In-Context LearningLlama3-8B Scenario 3: More layers (L={0,4,8,...})80.43Accuracy3In-Context LearningLlama3-8B Scenario 1: Diff. Pos. (P={4})74.1Accuracy3