Calibration Error Estimation on Amazon N=20,000
0.31MAE CWCE (Equal Binning 15)Equal Binning
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Equal BinningModel=Roberta2026.06 | 0.31 | 0.23 | — | — | — | — | |
| Equal BinningModel=Distill Bert2026.06 | 1.18 | 1.04 | — | — | — | — | |
| Equal BinningModel=Distill Roberta2026.06 | 1.65 | 1.43 | — | — | — | — | |
| Equal BinningModel=Bert2026.06 | 2.07 | 1.83 | — | — | — | — | |
| Adaptive BinningModel=Bert2026.06 | — | — | 4.83 | 3.38 | — | — | |
| Adaptive BinningModel=Distill Bert2026.06 | — | — | 3.93 | 2.57 | — | — | |
| Adaptive BinningModel=Roberta2026.06 | — | — | 3.8 | 2.27 | — | — | |
| Adaptive BinningModel=Distill Roberta2026.06 | — | — | 4.23 | 2.96 | — | — | |
| KDE MLEModel=Bert2026.06 | — | — | — | — | 0.85 | — | |
| KDE MLEModel=Distill Bert2026.06 | — | — | — | — | 0.99 | — | |
| KDE MLEModel=Roberta2026.06 | — | — | — | — | 0.95 | — | |
| KDE MLEModel=Distill Roberta2026.06 | — | — | — | — | 1.1 | — | |
| RAModel=Bert2026.06 | — | — | — | — | — | 0.38 | |
| RAModel=Distill Bert2026.06 | — | — | — | — | — | 0.23 | |
| RAModel=Roberta2026.06 | — | — | — | — | — | 0.27 | |
| RAModel=Distill Roberta2026.06 | — | — | — | — | — | 0.23 |