Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

MSCINLI

Benchmarks

Task NameDataset NameSOTA ResultTrend
CalibrationMSciNLI
Expected Calibration Error (ECE)3.75
22
Natural Language InferenceMSciNLI
ECE2.08
16
ClassificationMSciNLI
Expected Calibration Error (ECE)1.52
16
Natural Language InferenceMSciNLI
AUROC68.27
5
Scientific Natural Language InferenceMSCINLI (test)
Macro F1 (Hardware)77.41
4
Showing 5 of 5 rows