Uncertainty Quantification for Artistic Rendition Classification on ImageNet-R (test)
94.4Accuracy @ 90% RejectionREPVLM
Evaluation results
| Method | Links | ||
|---|---|---|---|
| REPVLMBackbone=SigLIP ViT-B/16, Pre-training Data=DataComp2026.01 | 94.4 | 1 | |
| REPVLMBackbone=SigLIP ViT-B/16, Pre-training Data=Conceptual Caption2026.01 | 94.2 | 1 | |
| REPVLMBackbone=SigLIP ViT-B/16, Pre-training Data=LAION2026.01 | 93.9 | 1 | |
| MCDOBackbone=SigLIP ViT-B/16, Pre-training Data=Conceptual Caption2026.01 | 92.9 | 0.995 | |
| MCDOBackbone=SigLIP ViT-B/16, Pre-training Data=DataComp2026.01 | 92.9 | 0.995 | |
| MCDOBackbone=SigLIP ViT-B/16, Pre-training Data=LAION2026.01 | 92.9 | 0.995 | |
| ProbVLMBackbone=SigLIP ViT-B/16, Pre-training Data=DataComp2026.01 | 91.9 | 0.995 | |
| ProbVLMBackbone=SigLIP ViT-B/16, Pre-training Data=Conceptual Caption2026.01 | 90.3 | 0.903 | |
| ProbVLMBackbone=SigLIP ViT-B/16, Pre-training Data=LAION2026.01 | 90.1 | 0.821 |