Object Hallucination Evaluation on MSCOCO 2014 (val)
56.8CHAIRsVDD-None
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| VDD-NoneBackbone=LLaVA-1.5-7B, Regime=Greedy2026.05 | 56.8 | 16.57 | — | 81.84 | |
| BaselineBackbone=LLaVA-1.5-7B, Regime=Nucleus2026.05 | 56.6 | 17.84 | — | 73.13 | |
| VDD-NoneBackbone=LLaVA-1.5-7B, Regime=Nucleus2026.05 | 56.6 | 17.04 | — | 76.04 | |
| VanillaDecoding=Greedy, Model=Shikra2025.12 | 54.6 | 15 | — | — | |
| BeamBackbone=LLaVA-1.5, Number of beams=52024.12 | 53.84 | 15.6 | 87.47 | — | |
| VanillaDecoding=Beam Search, Model=Shikra2025.12 | 53.4 | 14.1 | — | — | |
| VCDBackbone=LLaVA-1.52024.12 | 51.92 | 15.42 | 83.12 | — | |
| DoLaBackbone=LLaVA-1.52024.12 | 50.88 | 14.64 | 82.41 | — | |
| BaselineBackbone=LLaVA-1.5-7B, Regime=Beam2026.05 | 50.8 | 14.97 | — | 77.4 | |
| CODEBackbone=InstructBLIP2024.12 | 50.76 | 14.12 | 88.57 | — | |
| VCDBackbone=InstructBLIP2024.12 | 50.72 | 14.42 | 90.39 | — | |
| VCDDecoding=Beam Search, Model=LLAVA-1.52025.12 | 50 | 14.4 | — | — | |
| VanillaDecoding=Beam Search, Model=LLAVA-1.52025.12 | 49.8 | 14 | — | — | |
| GreedyBackbone=LLaVA-1.52024.12 | 49.68 | 14.32 | 83.06 | — | |
| BaselineBackbone=Qwen-VL-Chat, Regime=Nucleus2026.05 | 49.4 | 13.07 | — | 73.6 | |
| VCDBackbone=LLaVA-1.5-7B, Regime=Nucleus2026.05 | 49.2 | 14.79 | — | 75.34 | |
| PAIBackbone=Qwen-VL-Chat, Regime=Greedy2026.05 | 49 | 12.54 | — | 76.56 | |
| BeamBackbone=InstructBLIP, Number of beams=52024.12 | 48.56 | 13.5 | 94.87 | — | |
| PAIBackbone=LLaVA-1.5-7B, Regime=Beam2026.05 | 48.2 | 14.75 | — | 78.84 | |
| CODEBackbone=LLaVA-1.52024.12 | 47.96 | 14.26 | 78.52 | — | |
| BaselineBackbone=LLaVA-1.5-7B, Regime=Greedy2026.05 | 47.8 | 13.03 | — | 78.94 | |
| VCDBackbone=Qwen-VL-Chat, Regime=Nucleus2026.05 | 47.6 | 12.06 | — | 74.89 | |
| BRACSBackbone=LLaVA-1.5-7B, Regime=Nucleus2026.05 | 47.2 | 13.6 | — | 74.85 | |
| SPINBackbone=LLaVA-1.5-7B, Regime=Beam2026.05 | 46.6 | 14.2 | — | 73.54 | |
| BaselineBackbone=Qwen-VL-Chat, Regime=Greedy2026.05 | 46.6 | 12.52 | — | 75.6 | |
| EAHBackbone=InstructBLIP2024.12 | 46.4 | 13.13 | 92.33 | — | |
| DoLaBackbone=InstructBLIP2024.12 | 46 | 13 | 90.75 | — | |
| OPERABackbone=InstructBLIP, Number of beams=52024.12 | 45.76 | 13.06 | 92.46 | — | |
| BaselineBackbone=Qwen-VL-Chat, Regime=Beam2026.05 | 45.5 | 12.97 | — | 76.64 | |
| GreedyBackbone=InstructBLIP2024.12 | 45.32 | 12.98 | 91.06 | — | |
| BRACSBackbone=LLaVA-1.5-7B, Regime=Beam2026.05 | 44.8 | 13.52 | — | 76.66 | |
| VDD-NoneBackbone=Qwen-VL-Chat, Regime=Nucleus2026.05 | 44.4 | 11.55 | — | 76.11 | |
| PAIBackbone=LLaVA-1.5-7B, Regime=Greedy2026.05 | 44.3 | 12.58 | — | 75.03 | |
| OPERABackbone=LLaVA-1.5, Number of beams=52024.12 | 44.28 | 13.36 | 75.88 | — | |
| VanillaDecoding=Greedy, Model=LLAVA-1.52025.12 | 43.8 | 13 | — | — | |
| VCDDecoding=Greedy, Model=LLAVA-1.52025.12 | 43.5 | 13.8 | — | — | |
| SPINBackbone=LLaVA-1.5-7B, Regime=Greedy2026.05 | 43.2 | 14.42 | — | 73.65 | |
| BRACSBackbone=Qwen-VL-Chat, Regime=Nucleus2026.05 | 42.2 | 11.02 | — | 72.05 | |
| VDD-NoneBackbone=Qwen-VL-Chat, Regime=Greedy2026.05 | 41.4 | 10.43 | — | 72.76 | |
| SPINBackbone=Qwen-VL-Chat, Regime=Beam2026.05 | 41 | 11.53 | — | 75.84 | |
| BRACSBackbone=Qwen-VL-Chat, Regime=Greedy2026.05 | 40.8 | 11.17 | — | 77.4 | |
| PAIBackbone=Qwen-VL-Chat, Regime=Beam2026.05 | 40.2 | 11.08 | — | 75.36 | |
| BRACSBackbone=LLaVA-1.5-7B, Regime=Greedy2026.05 | 40 | 10.88 | — | 77.33 | |
| EAHBackbone=LLaVA-1.52024.12 | 38.76 | 11.05 | 86.28 | — | |
| BRACSBackbone=Qwen-VL-Chat, Regime=Beam2026.05 | 38 | 10.48 | — | 75.24 | |
| VHRBackbone=InstructBLIP2024.12 | 37.76 | 9.75 | 106.49 | — | |
| PAIDecoding=Beam Search, Model=Shikra2025.12 | 36.2 | 9.8 | — | — | |
| VanillaDecoding=Beam Search, Model=MiniGPT-42025.12 | 34.6 | 10.1 | — | — | |
| VanillaDecoding=Greedy, Model=MiniGPT-42025.12 | 33.8 | 10.4 | — | — | |
| VHRBackbone=LLaVA-1.52024.12 | 33.32 | 9.71 | 81.33 | — | |
| VISTADecoding=Beam Search, Model=Shikra2025.12 | 33 | 9.6 | — | — | |
| PAIDecoding=Beam Search, Model=MiniGPT-42025.12 | 31.8 | 9.9 | — | — | |
| PAIDecoding=Greedy, Model=Shikra2025.12 | 31.6 | 8.9 | — | — | |
| VISTADecoding=Greedy, Model=Shikra2025.12 | 31.4 | 8.6 | — | — | |
| SPINBackbone=Qwen-VL-Chat, Regime=Greedy2026.05 | 31 | 8.68 | — | 72.38 | |
| VCDBackbone=LLaVA-NeXT2024.12 | 30.8 | 8.72 | 157.72 | — | |
| [15]Decoding=Beam Search, Model=MiniGPT-42025.12 | 29.8 | 8.8 | — | — | |
| [15]Decoding=Greedy, Model=LLAVA-1.52025.12 | 29.5 | 8.8 | — | — | |
| [15]Decoding=Beam Search, Model=LLAVA-1.52025.12 | 29.4 | 8.5 | — | — | |
| GreedyBackbone=LLaVA-NeXT2024.12 | 29.08 | 8.08 | 157.06 | — | |
| PAIDecoding=Greedy, Model=LLAVA-1.52025.12 | 28.8 | 7.8 | — | — | |
| DoLaBackbone=LLaVA-NeXT2024.12 | 28.76 | 8.12 | 155.75 | — | |
| EAHBackbone=LLaVA-NeXT2024.12 | 28.13 | 6.62 | 142.75 | — | |
| CODEBackbone=LLaVA-NeXT2024.12 | 27.84 | 7.98 | 151.51 | — | |
| PAIDecoding=Beam Search, Model=LLAVA-1.52025.12 | 27.5 | 7.8 | — | — | |
| VISTADecoding=Greedy, Model=LLAVA-1.52025.12 | 27.2 | 7.3 | — | — | |
| BeamBackbone=LLaVA-NeXT, Number of beams=52024.12 | 25.72 | 6.92 | 160.64 | — | |
| VHRBackbone=LLaVA-NeXT2024.12 | 24.96 | 6.8 | 156.92 | — | |
| VEGASDecoding=Greedy, Model=LLAVA-1.52025.12 | 24.8 | 7.1 | — | — | |
| VEGASDecoding=Beam Search, Model=Shikra2025.12 | 24.5 | 8.8 | — | — | |
| PAIDecoding=Greedy, Model=MiniGPT-42025.12 | 24 | 9 | — | — | |
| VEGASDecoding=Greedy, Model=Shikra2025.12 | 24 | 8.1 | — | — | |
| VISTADecoding=Beam Search, Model=LLAVA-1.52025.12 | 24 | 7.5 | — | — | |
| [15]Decoding=Beam Search, Model=Shikra2025.12 | 23.5 | 9.1 | — | — | |
| [15]Decoding=Greedy, Model=MiniGPT-42025.12 | 23.4 | 8.8 | — | — | |
| [15]Decoding=Greedy, Model=Shikra2025.12 | 23.4 | 8.1 | — | — | |
| VEGASDecoding=Beam Search, Model=LLAVA-1.52025.12 | 23.2 | 7 | — | — | |
| VISTADecoding=Beam Search, Model=MiniGPT-42025.12 | 22.8 | 8.2 | — | — | |
| VISTADecoding=Greedy, Model=MiniGPT-42025.12 | 22.5 | 8.7 | — | — | |
| VEGASDecoding=Beam Search, Model=MiniGPT-42025.12 | 21.7 | 8 | — | — | |
| VEGASDecoding=Greedy, Model=MiniGPT-42025.12 | 21.4 | 8.4 | — | — |