Hallucination Evaluation on MME Count
70MME Count (Zh)Baseline
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| BaselineBackbone=LLaVA-1.52025.06 | 70 | 155 | 55 | 58.3 | 80 | 85 | — | — | 69.67 | |
| VCDBackbone=LLaVA-1.52025.06 | 73.3 | — | 73.3 | 53.3 | 61.7 | 105 | — | — | 73.33 | |
| CLAIMBackbone=LLaVA-1.52025.06 | 130 | 125 | 110 | 130 | 135 | — | — | — | 126 | |
| BaselineBackbone=Qwen-VL-Chat2025.06 | 130 | 150 | 143.3 | 113.3 | — | — | 60 | 136.7 | 116.7 | |
| VCDBackbone=Qwen-VL-Chat2025.06 | 140 | — | 131.7 | 100 | — | — | 80 | 128.3 | 116 | |
| CLAIMBackbone=Qwen-VL-Chat2025.06 | 153.3 | 148.3 | 120 | 111.7 | — | — | 137 | — | 134 |