Image Classification on CIFAR10 (Calibration and Uncertainty)
88.2AUROCSGPU
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| SGPUEvaluation=Averaged across LVLM architectures2025.12 | 88.2 | 87.8 | 3.6 | |
| Cos EigenscoreEvaluation=Averaged across LVLM architectures2025.12 | 84.1 | 77 | 12.7 | |
| KLE-MaternEvaluation=Averaged across LVLM architectures2025.12 | 83.2 | 83.7 | 6.6 | |
| KLE-HeatEvaluation=Averaged across LVLM architectures2025.12 | 82.7 | 83.8 | 8.4 | |
| UMPIREEvaluation=Averaged across LVLM architectures2025.12 | 82.6 | 80 | 12.2 | |
| SEEvaluation=Averaged across LVLM architectures2025.12 | 82.3 | 83.2 | 7.5 | |
| DSEEvaluation=Averaged across LVLM architectures2025.12 | 81.8 | 83.2 | 7 | |
| PEEvaluation=Averaged across LVLM architectures2025.12 | 81.2 | 82.4 | 8.1 | |
| Cov EigenscoreEvaluation=Averaged across LVLM architectures2025.12 | 78.7 | 75 | 24.3 |