Multimodal Spatial Reasoning on VSR
70.14AccuracySpecFlow
Evaluation Results
| Method | Links | |
|---|---|---|
| SpecFlowFLOPs(G)=11169.74, Lat.(s)=0.41, Mem.(GB)=39.532026.06 | 70.14 | |
| VoCoTFLOPs(G)=20342.40, Lat.(s)=0.65, Mem.(GB)=56.372026.06 | 68.88 | |
| CODIFLOPs(G)=11099.76, Lat.(s)=0.46, Mem.(GB)=39.022026.06 | 68.82 | |
| SparseVLMFLOPs(G)=15200.29, Lat.(s)=0.50, Mem.(GB)=45.752026.06 | 63.88 | |
| LightFastVFLOPs(G)=16280.49, Lat.(s)=0.55, Mem.(GB)=49.652026.06 | 58.63 | |
| PCCoTFLOPs(G)=10654.00, Lat.(s)=0.42, Mem.(GB)=39.792026.06 | 54.71 | |
| SoTFLOPs(G)=17675.04, Lat.(s)=0.59, Mem.(GB)=52.622026.06 | 54.56 | |
| HeimaFLOPs(G)=10394.46, Lat.(s)=0.40, Mem.(GB)=38.842026.06 | 51.69 |