Classification on 16 datasets average
61.79AccuracyTTE
Evaluation results
| Method | Links | ||
|---|---|---|---|
| TTE2025.03 | 61.79 | 7.79 | |
| RNDescription=Random Noise (N=0 special case of TTC)2025.03 | 61.61 | 0.06 | |
| CLIP2025.03 | 61.51 | 0.09 | |
| Anti-adv2025.03 | 57.32 | 0.53 | |
| HD2025.03 | 56.62 | 1.19 | |
| TTC2025.03 | 55.99 | 20.63 | |
| CLIP-FTEvaluation Protocol=Fine-tuned2025.03 | 55.8 | 0.96 | |
| FAREEvaluation Protocol=Fine-tuned, Finetuning Attack Budget=1/2552025.03 | 51.02 | 1.5 | |
| FAREEvaluation Protocol=Fine-tuned, Finetuning Attack Budget=4/2552025.03 | 46.17 | 3.67 | |
| ARGENTModel Size=Large2026.03 | 45.1 | — | |
| HyCoCLIPModel Size=Large2026.03 | 44.4 | — | |
| ARGENTModel Size=Base2026.03 | 42.8 | — | |
| HyCoCLIPModel Size=Base2026.03 | 42.6 | — | |
| PMG-AFTEvaluation Protocol=Fine-tuned, Finetuning Attack Budget=1/2552025.03 | 42.3 | 7.03 | |
| CLIPModel Size=Large2026.03 | 40.6 | — | |
| TeCoAEvaluation Protocol=Fine-tuned, Finetuning Attack Budget=1/2552025.03 | 40.25 | 6.51 | |
| MERUModel Size=Large2026.03 | 40.2 | — | |
| HyCoCLIPModel Size=Small2026.03 | 39.2 | — | |
| ARGENTModel Size=Small2026.03 | 38.4 | — | |
| CLIPModel Size=Base2026.03 | 37.6 | — | |
| PMG-AFTEvaluation Protocol=Fine-tuned, Finetuning Attack Budget=4/2552025.03 | 37.58 | 10.7 | |
| MERUModel Size=Base2026.03 | 37 | — | |
| TeCoAEvaluation Protocol=Fine-tuned, Finetuning Attack Budget=4/2552025.03 | 35.57 | 10.03 | |
| CLIPModel Size=Small2026.03 | 35.5 | — | |
| MERUModel Size=Small2026.03 | 34.3 | — |