Self-evaluation on VisualCoT
80.2AUROCVAUQ
Evaluation results
| Method | Links | |
|---|---|---|
| VAUQModel=LLaVA-1.5-13B2026.02 | 80.2 | |
| VAUQModel=LLaVA-1.5-7B2026.02 | 77.8 | |
| VL-UncertaintyModel=LLaVA-1.5-13B2026.02 | 77.7 | |
| VAUQBackbone=InternVL3.5-8B2026.02 | 77.2 | |
| Semantic EntropyModel=LLaVA-1.5-13B2026.02 | 75.7 | |
| EigenScoreModel=LLaVA-1.5-13B2026.02 | 75.5 | |
| Semantic EntropyModel=LLaVA-1.5-7B2026.02 | 75.1 | |
| EigenScoreModel=LLaVA-1.5-7B2026.02 | 74.7 | |
| EigenScoreBackbone=InternVL3.5-8B2026.02 | 74.5 | |
| VAUQBackbone=Qwen-2.5-VL-7B2026.02 | 68 | |
| Semantic EntropyBackbone=InternVL3.5-8B2026.02 | 66.3 | |
| VL-UncertaintyBackbone=InternVL3.5-8B2026.02 | 65.8 | |
| VL-UncertaintyModel=LLaVA-1.5-7B2026.02 | 65.2 | |
| Contextual LensBackbone=Qwen-2.5-VL-7B2026.02 | 65 | |
| PerplexityModel=LLaVA-1.5-13B2026.02 | 63.7 | |
| VL-UncertaintyBackbone=Qwen-2.5-VL-7B2026.02 | 62.3 | |
| PerplexityBackbone=InternVL3.5-8B2026.02 | 62.3 | |
| Chain-of-EmbeddingsBackbone=InternVL3.5-8B2026.02 | 61.4 | |
| Contextual LensModel=LLaVA-1.5-7B2026.02 | 58.3 | |
| SVARBackbone=InternVL3.5-8B2026.02 | 56.7 | |
| PerplexityModel=LLaVA-1.5-7B2026.02 | 56.2 | |
| PerplexityBackbone=Qwen-2.5-VL-7B2026.02 | 56 | |
| Chain-of-EmbeddingsModel=LLaVA-1.5-7B2026.02 | 55.7 | |
| Contextual LensBackbone=InternVL3.5-8B2026.02 | 55.1 | |
| VerbalizedBackbone=Qwen-2.5-VL-7B2026.02 | 54.7 | |
| VerbalizedBackbone=InternVL3.5-8B2026.02 | 54.7 | |
| VerbalizedModel=LLaVA-1.5-13B2026.02 | 54.4 | |
| Semantic EntropyBackbone=Qwen-2.5-VL-7B2026.02 | 53.3 | |
| Contextual LensModel=LLaVA-1.5-13B2026.02 | 52.5 | |
| EigenScoreBackbone=Qwen-2.5-VL-7B2026.02 | 51.1 | |
| VerbalizedModel=LLaVA-1.5-7B2026.02 | 49.9 | |
| SVARBackbone=Qwen-2.5-VL-7B2026.02 | 46.7 | |
| Chain-of-EmbeddingsModel=LLaVA-1.5-13B2026.02 | 46.5 | |
| SVARModel=LLaVA-1.5-7B2026.02 | 44.5 | |
| SVARModel=LLaVA-1.5-13B2026.02 | 44.5 | |
| Chain-of-EmbeddingsBackbone=Qwen-2.5-VL-7B2026.02 | 44.3 |