Weight-space classification on Model Jungle (accuracy)
92.33AccuracyMVProbe
Evaluation results
| Method | Links | |
|---|---|---|
| MVProbeBackbone=SupViT2026.05 | 92.33 | |
| MVProbeBackbone=ResNet2026.05 | 92.24 | |
| ProbeX (×4)Backbone=SupViT2026.05 | 90.33 | |
| ProbeXBackbone=SupViT2026.05 | 88.08 | |
| ProbeX (×4)Backbone=ResNet2026.05 | 87.16 | |
| MVProbeBackbone=MAE2026.05 | 81.62 | |
| ProbeXBackbone=ResNet2026.05 | 81.61 | |
| ProbeGenBackbone=SupViT2026.05 | 78.48 | |
| MVProbeBackbone=DINO2026.05 | 78.29 | |
| ProbeGenBackbone=ResNet2026.05 | 78.27 | |
| ProbeX (×4)Backbone=MAE2026.05 | 77.26 | |
| ProbeXBackbone=MAE2026.05 | 77.11 | |
| ProbeX (×4)Backbone=DINO2026.05 | 73.25 | |
| ProbeXBackbone=DINO2026.05 | 72.54 | |
| ProbeGenBackbone=MAE2026.05 | 70.68 | |
| ProbeGenBackbone=DINO2026.05 | 61.26 | |
| StatNNBackbone=SupViT2026.05 | 55.8 | |
| StatNNBackbone=DINO2026.05 | 55.69 | |
| StatNNBackbone=ResNet2026.05 | 55.2 | |
| StatNNBackbone=MAE2026.05 | 54.83 |