Visual Question Answering on ADVQA
71.1AUROCSGPU
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SGPUEvaluation=Averaged across LVLM architectures2025.12 | 71.1 | 70.1 | 0.173 | |
| KLE-MaternEvaluation=Averaged across LVLM architectures2025.12 | 70 | 67.2 | 0.105 | |
| KLE-HeatEvaluation=Averaged across LVLM architectures2025.12 | 68.8 | 66.7 | 0.168 | |
| DSEEvaluation=Averaged across LVLM architectures2025.12 | 67.6 | 65.9 | 0.152 | |
| SEEvaluation=Averaged across LVLM architectures2025.12 | 67.3 | 65.8 | 0.149 | |
| PEEvaluation=Averaged across LVLM architectures2025.12 | 67.2 | 65.7 | 0.168 | |
| UMPIREEvaluation=Averaged across LVLM architectures2025.12 | 66.7 | 62.1 | 0.229 | |
| Cos EigenscoreEvaluation=Averaged across LVLM architectures2025.12 | 63.5 | 58.7 | 0.201 | |
| Cov EigenscoreEvaluation=Averaged across LVLM architectures2025.12 | 62 | 58.1 | 0.324 |