Image Classification on ImageNet-1K (test) (Acc @ 90% Rej & S Score)
84.7Acc @ 90% RejREPVLM
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| REPVLMBackbone=SigLIP ViT-B/16, Pre-training Data=DataComp2026.01 | 84.7 | — | 0.905 | |
| REPVLMBackbone=SigLIP ViT-B/16, Pre-training Data=Conceptual Caption2026.01 | 82.4 | — | 0.91 | |
| REPVLMBackbone=SigLIP ViT-B/16, Pre-training Data=LAION2026.01 | 82 | — | 0.985 | |
| ProbVLMBackbone=SigLIP ViT-B/16, Pre-training Data=DataComp2026.01 | 77.9 | — | 0.932 | |
| MCDOBackbone=SigLIP ViT-B/16, Pre-training Data=Conceptual Caption2026.01 | 75.9 | — | -0.362 | |
| MCDOBackbone=SigLIP ViT-B/16, Pre-training Data=DataComp2026.01 | 75.9 | — | -0.362 | |
| MCDOBackbone=SigLIP ViT-B/16, Pre-training Data=LAION2026.01 | 75.9 | — | -0.362 | |
| ProbVLMBackbone=SigLIP ViT-B/16, Pre-training Data=LAION2026.01 | 73.7 | — | -0.947 | |
| REPVLMTraining configuration=LAION2026.01 | 73.4 | 0.998 | — | |
| REPVLMTraining configuration=Conceptual Caption2026.01 | 73 | 0.993 | — | |
| REPVLMTraining configuration=DataComp2026.01 | 72.3 | 1 | — | |
| ProbVLMBackbone=SigLIP ViT-B/16, Pre-training Data=Conceptual Caption2026.01 | 71.2 | — | -1 | |
| MCDOTraining configuration=Conceptual Caption2026.01 | 69.9 | 0.99 | — | |
| MCDOTraining configuration=DataComp2026.01 | 69.9 | 0.99 | — | |
| MCDOTraining configuration=LAION2026.01 | 69.9 | 0.99 | — | |
| ProbVLMTraining configuration=DataComp2026.01 | 67 | 0.435 | — | |
| ProbVLMTraining configuration=LAION2026.01 | 66.7 | 0.202 | — | |
| ProbVLMTraining configuration=Conceptual Caption2026.01 | 66.4 | 0.423 | — |