Multi-class Calibration on IMAGENET (test)
-0.049NLL Improvement (Absolute)SVS
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| SVSModel=resnet1522025.11 | -0.049 | — | — | — | — | — | |
| SVSBackbone=resnet1522025.11 | -0.049 | — | — | — | — | — | |
| TS probm.Model=resnet1522025.11 | -0.047 | — | — | — | — | — | |
| TS probmetricsBackbone=resnet1522025.11 | -0.047 | — | — | — | — | — | |
| SVSModel=densenet1612025.11 | -0.037 | — | — | — | — | — | |
| SVSBackbone=densenet1612025.11 | -0.037 | — | — | — | — | — | |
| TS probm.Model=densenet1612025.11 | -0.035 | — | — | — | — | — | |
| TS probmetricsBackbone=densenet1612025.11 | -0.035 | — | — | — | — | — | |
| torchcal VSModel=resnet1522025.11 | -0.015 | — | — | — | — | — | |
| VS torchcalBackbone=resnet1522025.11 | -0.015 | — | — | — | — | — | |
| torchcal VSModel=densenet1612025.11 | 0.007 | — | — | — | — | — | |
| VS torchcalBackbone=densenet1612025.11 | 0.007 | — | — | — | — | — | |
| FIRType=Average Ranking2025.12 | — | 2 | 3 | 2.4 | 3.1 | 2.5 | |
| IR OVRType=Average Ranking2025.12 | — | 8 | 5.5 | 6.5 | 4.4 | 8 | |
| MSType=Average Ranking2025.12 | — | 5.9 | 7.4 | 6.6 | 5.4 | 6.1 | |
| NA-FIRType=Average Ranking2025.12 | — | 1 | 1.4 | 1.2 | 3.6 | 2.9 | |
| SCIRType=Average Ranking2025.12 | — | 4.9 | 2 | 3.4 | 6.1 | 2.1 | |
| TSType=Average Ranking2025.12 | — | 3.4 | 4 | 4.1 | 2 | 5.1 | |
| UNCALIBRATEDType=Average Ranking2025.12 | — | 5.8 | 6.6 | 6 | 7.2 | 2.9 | |
| VSType=Average Ranking2025.12 | — | 5.1 | 6.1 | 5.8 | 4.1 | 6.4 |