ResearchBenchmarksFailure Detection and Reasoning on ARMBench S→AFailure Detection and Reasoning on ARMBench S→AFollow72.5Detection AccuracyARMOR48.5854.796167.21Feb 12, 2026Evaluation resultsMethodMethodLinksDetection AccuracyLLM Fuzzy ScoreROUGELARMORdomain_transfer=Sparro...domain_transfer=Sparrow to ARMBench2026.0272.50.69872.1SFT-S+Ddomain_transfer=Sparro...domain_transfer=Sparrow to ARMBench2026.0249.50.00724.9