Visual Question Answering on OKVQA (AUROC, AUARC, ECE)
0.788AUROCSGPU
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SGPUEvaluation=Averaged across LVLM architectures2025.12 | 0.788 | 0.81 | 0.072 | |
| UMPIREEvaluation=Averaged across LVLM architectures2025.12 | 0.768 | 0.777 | 0.236 | |
| SEEvaluation=Averaged across LVLM architectures2025.12 | 0.763 | 0.792 | 0.232 | |
| KLE-MaternEvaluation=Averaged across LVLM architectures2025.12 | 0.762 | 0.794 | 0.075 | |
| DSEEvaluation=Averaged across LVLM architectures2025.12 | 0.753 | 0.787 | 0.213 | |
| Cos EigenscoreEvaluation=Averaged across LVLM architectures2025.12 | 0.75 | 0.738 | 0.123 | |
| PEEvaluation=Averaged across LVLM architectures2025.12 | 0.737 | 0.783 | 0.109 | |
| KLE-HeatEvaluation=Averaged across LVLM architectures2025.12 | 0.731 | 0.782 | 0.191 | |
| Cov EigenscoreEvaluation=Averaged across LVLM architectures2025.12 | 0.685 | 0.705 | 0.28 |