Classification Calibration on WiDS
16Delta NLL (%)Clustered Calibration
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Clustered CalibrationModel Capacity=Mean across 6 capacities, Comparison Baseline=Best global parametric calibrator2025.10 | 16 | 10 | 9 | 11.59 |
Loading the SOTA2 catalog…