Hallucination Evaluation on CPBench
78.51CPQAntidote
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| AntidoteBackbone=LLaVA-1.5-7B, Hyperparameter Configuration=HP-search2026.04 | 78.51 | 70.54 | 73.47 | |
| HalluVL-DPOwBackbone=Qwen2-VL-7B2026.04 | 75 | 69.59 | 71.52 | |
| HalluVL-DPOwBackbone=LLaVA-1.5-7B2026.04 | 73.51 | 72.43 | 72.82 | |
| HalluVL-DPOBackbone=LLaVA-1.5-7B2026.04 | 71.22 | 75.54 | 73.94 | |
| HalluVL-DPOBackbone=Qwen2-VL-7B2026.04 | 63.38 | 75.27 | 71.05 | |
| AntidoteBackbone=LLaVA-1.5-7B, Hyperparameter Configuration=HP-orig2026.04 | 27.7 | 90.27 | 68.76 | |
| VCDBackbone=Qwen2-VL-7B2026.04 | 23.78 | 88.36 | 66.79 | |
| Qwen2-VL-7BBackbone=Qwen2-VL-7B2026.04 | 19.19 | 87.97 | 65.46 | |
| VCDBackbone=LLaVA-1.5-7B2026.04 | 1.76 | 100 | 67.06 | |
| LLaVA-1.5-7BBackbone=LLaVA-1.5-7B2026.04 | 0.54 | 100 | 66.79 |