List your company

Selective Prediction on GSM8K 60 (test)

0.994AUROCRidge Atlas
0.622720.719110.81550.91189Jun 8, 2026

Evaluation results

MethodLinks
Ridge AtlasModel=Gemma-2-9B-IT, Variant=TRiE/Ridge Proximity2026.060.9940.00299.80.2
Baseline kNNModel=Gemma-2-9B-IT, Variant=kNN-R² proximity2026.060.8130.01298.81
Neg. H0 total persistenceModel=Gemma-2-9B-IT2026.060.6370.02497.51.8