Image Classification on ImageNet-to-Target Generalization Suite τ=0.25 (test)
69.89IN AccuracyCPT
Evaluation Results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CPTImbalance Ratio (τ)=0.25, Classification Protocol=Base-to-New2026.05 | 69.89 | 94.56 | 91.27 | 66.64 | 73.09 | 87.03 | 23.02 | 67.54 | 46.53 | 50.16 | 67.82 | 66.76 | |
| DeKgImbalance Ratio (τ)=0.25, Classification Protocol=Base-to-New2026.05 | 69.46 | 93.79 | 90.02 | 65.49 | 72.39 | 86.59 | 22.6 | 66.96 | 44.57 | 49.96 | 67.33 | 65.97 | |
| NPTImbalance Ratio (τ)=0.25, Classification Protocol=Base-to-New2026.05 | 69.4 | 93.6 | 90.4 | 65.95 | 72.33 | 86.19 | 22.78 | 66.89 | 46.05 | 49.67 | 67.12 | 66.1 | |
| CoPromptImbalance Ratio (τ)=0.25, Classification Protocol=Base-to-New2026.05 | 69.34 | 92.64 | 89.76 | 65.27 | 71.89 | 85.35 | 22.54 | 66.24 | 45.58 | 49.2 | 66.42 | 65.49 | |
| DPCImbalance Ratio (τ)=0.25, Classification Protocol=Base-to-New2026.05 | 69.21 | 92.63 | 89.96 | 64.65 | 70.36 | 85.33 | 21.44 | 65.75 | 45.6 | 45.5 | 65.76 | 64.7 | |
| MaPLeImbalance Ratio (τ)=0.25, Classification Protocol=Base-to-New2026.05 | 69.2 | 92.63 | 90.34 | 65.17 | 71.54 | 85.03 | 21.39 | 65.57 | 45.84 | 47.83 | 65.34 | 65.07 |