Image Classification on ImageNet H
73.93Top-1 AccuracyHystar
Evaluation results
| Method | Links | |
|---|---|---|
| HystarShot=16-shot, Protocol=Base-to-new generalization, Backbone=CLIP2026.05 | 73.93 | |
| MaPLeShot=16-shot, Protocol=Base-to-new generalization, Backbone=CLIP2026.05 | 73.47 | |
| TCPShot=16-shot, Protocol=Base-to-new generalization, Backbone=CLIP2026.05 | 73.38 | |
| KgCoOpShot=16-shot, Protocol=Base-to-new generalization, Backbone=CLIP2026.05 | 72.78 | |
| ProGradShot=16-shot, Protocol=Base-to-new generalization, Backbone=CLIP2026.05 | 71.46 | |
| CoOpShot=16-shot, Protocol=Base-to-new generalization, Backbone=CLIP2026.05 | 71.02 | |
| DREAMEvaluation Protocol=Fine-tuning2026.03 | 26 | |
| CLIPEvaluation Protocol=Fine-tuning2026.03 | 23.3 | |
| REPAEvaluation Protocol=Fine-tuning2026.03 | 23.1 | |
| MAREvaluation Protocol=Fine-tuning2026.03 | 21.1 | |
| FLUIDEvaluation Protocol=Fine-tuning2026.03 | 21.1 | |
| NS-WSBackbone=ResNet-18, Optimization Strategy=NS-WS2026.04 | 10.2 | |
| ResNet-18Backbone=ResNet-18, Optimization Strategy=Baseline2026.04 | 9.9 |