ResearchBenchmarksSelective Prediction on MNLIFollow86.92Coverage @ 0.9UAT-LITE70.768874.961979.15583.3481Feb 3, 2026Evaluation ResultsMethodMethodLinksCoverage @ 0.9Coverage @ 0.8Coverage @ 0.7AURCUAT-LITEevaluation=mean over f...evaluation=mean over five seeds2026.0286.9290.392.750.0529Baselinetype=deterministictype=deterministic2026.0271.3975.2779.260.164