Near Out-of-Distribution Detection on 13 Near-OOD Datasets Average (test)
83.85MaxLogit AUROCPromptSRC
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| PromptSRCVisual Encoder=ViT-B/16, Text Encoder=Transformer, Few-shot training settings=1, 2, 4, 8, 16-shot average2024.05 | 83.85 | 85.77 | 1.92 | |
| IVLPVisual Encoder=ViT-B/16, Text Encoder=Transformer, Few-shot training settings=1, 2, 4, 8, 16-shot average2024.05 | 81.12 | 84.34 | 3.23 | |
| CoCoOpVisual Encoder=ViT-B/16, Text Encoder=Transformer, Few-shot training settings=1, 2, 4, 8, 16-shot average2024.05 | 81.09 | 82.74 | 1.65 | |
| MaPLeVisual Encoder=ViT-B/16, Text Encoder=Transformer, Few-shot training settings=1, 2, 4, 8, 16-shot average2024.05 | 81.06 | 83.94 | 2.88 | |
| KgCoOpVisual Encoder=ViT-B/16, Text Encoder=Transformer, Few-shot training settings=1, 2, 4, 8, 16-shot average2024.05 | 80.84 | 83.12 | 2.28 | |
| CoOpVisual Encoder=ViT-B/16, Text Encoder=Transformer, Few-shot training settings=1, 2, 4, 8, 16-shot average2024.05 | 80.74 | 81.84 | 1.09 | |
| ProGradVisual Encoder=ViT-B/16, Text Encoder=Transformer, Few-shot training settings=1, 2, 4, 8, 16-shot average2024.05 | 79.77 | 82.35 | 2.58 | |
| LoCoOpVisual Encoder=ViT-B/16, Text Encoder=Transformer, Few-shot training settings=1, 2, 4, 8, 16-shot average2024.05 | 77.55 | 81.74 | 4.18 |