Failure Detection and Reasoning on ARMBench S→A benchmark leaderboard · SOTA2 Research

Failure Detection and Reasoning on ARMBench S→A

72.5Detection AccuracyARMOR

Evaluation Results

Method	Links
ARMOR2026.02		72.5	0.698	72.1
SFT-S+D2026.02		49.5	0.007	24.9