OOD Detection on ImageNet-R (test)
88.84AUROCMM++
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MM++Backbone=ViT-B/16, ID Dataset=ImageNet-LT2026.06 | 88.84 | 41.83 | |
| X-Maha†Backbone=ViT-B/16, ID Dataset=ImageNet-LT2026.06 | 88.49 | 42.85 | |
| Maha++Backbone=ViT-B/16, ID Dataset=ImageNet-LT2026.06 | 87.22 | 46.92 | |
| MahaBackbone=ViT-B/16, ID Dataset=ImageNet-LT2026.06 | 86.23 | 51.85 | |
| rMaha++Backbone=ViT-B/16, ID Dataset=ImageNet-LT2026.06 | 85.93 | 49.18 | |
| rMahaBackbone=ViT-B/16, ID Dataset=ImageNet-LT2026.06 | 85.02 | 52 | |
| ReActBackbone=ViT-B/16, ID Dataset=ImageNet-LT2026.06 | 78.19 | — | |
| MSPBackbone=ViT-B/16, ID Dataset=ImageNet-LT2026.06 | 77.63 | 66.23 | |
| EnergyBackbone=ViT-B/16, ID Dataset=ImageNet-LT2026.06 | 72.87 | 67.04 | |
| ODINBackbone=ViT-B/16, ID Dataset=ImageNet-LT2026.06 | 67.9 | 74.49 | |
| KNNBackbone=ViT-B/16, ID Dataset=ImageNet-LT2026.06 | 65.61 | 84.17 |