Hallucination Evaluation on MME hallucination (test)
180Existence ScoreVASparse
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| VASparseBackbone=LLaVA-1.52025.01 | 180 | 132.67 | 121.33 | 160 | — | |
| VCDBackbone=LLaVA-1.52025.01 | 175.33 | 130.33 | 115 | 155 | — | |
| VASparseBackbone=mPLUG-Owl22025.01 | 175 | 130 | 110.67 | 155 | — | |
| VCDBackbone=mPLUG-Owl22025.01 | 171.33 | 125 | 107.33 | 150 | — | |
| DoLaBackbone=LLaVA-1.52025.01 | 170 | 120 | 106.67 | 150.67 | — | |
| HALCBackbone=LLaVA-1.5, Post-hoc=true2025.01 | 167.67 | 121.33 | 106.67 | 150.67 | — | |
| GreedyBackbone=mPLUG-Owl22025.01 | 167 | 120 | 105 | 145 | — | |
| DoLaBackbone=mPLUG-Owl22025.01 | 167 | 125 | 110 | 147.67 | — | |
| OPERABackbone=mPLUG-Owl22025.01 | 167 | 122.33 | 100 | 145 | — | |
| HALCBackbone=mPLUG-Owl2, Post-hoc=true2025.01 | 167 | 120.33 | 108.67 | 145 | — | |
| GreedyBackbone=LLaVA-1.52025.01 | 165.67 | 120 | 110.67 | 148.33 | — | |
| OPERABackbone=LLaVA-1.52025.01 | 165 | 115.67 | 104 | 145 | — | |
| VASparseBackbone=MiniGPT-42025.01 | 147.33 | 98.67 | 78.67 | 133 | — | |
| VCDBackbone=MiniGPT-42025.01 | 142 | 95.33 | 71.33 | 129 | — | |
| OPERABackbone=MiniGPT-42025.01 | 140.67 | 92.33 | 73 | 125 | — | |
| HALCBackbone=MiniGPT-4, Post-hoc=true2025.01 | 140 | 92.67 | 71.33 | 122.67 | — | |
| GreedyBackbone=MiniGPT-42025.01 | 137 | 93 | 75 | 125 | — | |
| DoLaBackbone=MiniGPT-42025.01 | 137 | 90 | 75.33 | 122.67 | — | |
| InstructBLIP + ICDBackbone=InstructBLIP, Method=ICD2024.03 | 136.67 | 90 | 76.67 | 123.33 | 426.67 | |
| InstructBLIP (default)Backbone=InstructBLIP, Method=default2024.03 | 135 | 53.33 | 56.67 | 93.33 | 338.33 | |
| InstructBLIP + VCDBackbone=InstructBLIP, Method=VCD2024.03 | 123.33 | 81.67 | 55 | 106.67 | 366.67 | |
| miniGPT4 + ICDBackbone=miniGPT4, Method=ICD2024.03 | 66.67 | 61.67 | 40 | 61.67 | 230.01 | |
| miniGPT4 + VCDBackbone=miniGPT4, Method=VCD2024.03 | 48.33 | 31.67 | 40 | 45 | 165 | |
| miniGPT4 (default)Backbone=miniGPT4, Method=default2024.03 | 46.67 | 26.67 | 38.33 | 38.33 | 150 |