Image Classification on ImageNet (Base/Novel Accuracy)
78.07Base AccuracyTransAgent
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| TransAgentVision Encoder=ViT-B/16, Evaluation Protocol=Base-to-novel generalization2024.10 | 78.07 | 70.57 | 74.13 | |
| PromptSRCVision Encoder=ViT-B/16, Evaluation Protocol=Base-to-novel generalization2024.10 | 77.6 | 70.73 | 74.01 | |
| RPOVision Encoder=ViT-B/16, Evaluation Protocol=Base-to-novel generalization2024.10 | 76.6 | 71.57 | 74 | |
| CoOpVision Encoder=ViT-B/16, Evaluation Protocol=Base-to-novel generalization2024.10 | 76.47 | 67.88 | 71.92 | |
| CoCoOpVision Encoder=ViT-B/16, Evaluation Protocol=Base-to-novel generalization2024.10 | 75.98 | 70.43 | 73.1 | |
| MaPLeVision Encoder=ViT-B/16, Evaluation Protocol=Base-to-novel generalization2024.10 | 75.4 | 70.32 | 72.72 | |
| CLIPVision Encoder=ViT-B/16, Evaluation Protocol=Base-to-novel generalization2024.10 | 72.43 | 68.14 | 70.22 |