Image Classification on Cars, EuroSAT, RESISC45, GTSRB (Corrupted Test Average)
85Accuracy (Motion Blur)DOGE AM
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| DOGE AMBackbone=ViT-B/32, Method Category=Test-Time Adaptation2025.08 | 85 | 65.7 | 73.4 | 44.2 | 75.4 | 83.8 | 84.2 | 73.1 | |
| RegMean++Backbone=ViT-B/32, Method Category=Training-Free2025.08 | 81.8 | 70 | 68.4 | 37.9 | 75.9 | 79.6 | 82.8 | 70.9 | |
| Layer-wise AdaMergingBackbone=ViT-B/32, Method Category=Test-Time Adaptation2025.08 | 81.3 | 69 | 71.2 | 41.3 | 74.5 | 80.2 | 80.1 | 71.1 | |
| RegMeanBackbone=ViT-B/32, Method Category=Training-Free2025.08 | 79.7 | 69.1 | 67.3 | 37.1 | 75.2 | 78 | 80.9 | 69.6 | |
| TSV-MBackbone=ViT-B/32, Method Category=Data-Free2025.08 | 78.9 | 69.9 | 69.1 | 37.8 | 75.4 | 77.2 | 80.2 | 69.8 | |
| DOGE TABackbone=ViT-B/32, Method Category=Data-Free2025.08 | 77.3 | 66 | 66.2 | 37.5 | 71.5 | 76.1 | 77.7 | 67.5 | |
| Iso-CBackbone=ViT-B/32, Method Category=Data-Free2025.08 | 75.9 | 62.2 | 63.9 | 35 | 69.2 | 76.5 | 75.3 | 65.4 | |
| Iso-CTSBackbone=ViT-B/32, Method Category=Data-Free2025.08 | 75.7 | 61.9 | 63.5 | 34.2 | 69.2 | 76.5 | 75.4 | 65.2 | |
| Fisher MergingBackbone=ViT-B/32, Method Category=Training-Free2025.08 | 67 | 59.8 | 60.7 | 29.3 | 64.9 | 67.9 | 69 | 59.8 | |
| Task ArithmeticBackbone=ViT-B/32, Method Category=Data-Free2025.08 | 65.9 | 58.9 | 59.6 | 29.7 | 63.5 | 66 | 67.8 | 58.8 | |
| Model SoupsBackbone=ViT-B/32, Method Category=Data-Free2025.08 | 64.6 | 56.9 | 58.1 | 28.5 | 61.3 | 64.7 | 66.3 | 57.2 | |
| TIES-MergingBackbone=ViT-B/32, Method Category=Data-Free2025.08 | 63.2 | 54.5 | 56.2 | 28.1 | 57.7 | 63.8 | 64.4 | 55.4 |