Object Probing on POPE Adversarial
83.9AccuracyAIR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AIRBase Model=LLaVA-1.52026.05 | 83.9 | 83.6 | |
| NoLanBase Model=LLaVA-1.52026.05 | 83.83 | 82.58 | |
| LaSCDBase Model=LLaVA-1.52026.05 | 83.46 | 82.6 | |
| DMASBase Model=LLaVA-1.52026.05 | 83.07 | 83.33 | |
| ASCDBase Model=LLaVA-1.52026.05 | 82.9 | 82.97 | |
| SIDBase Model=LLaVA-1.52026.05 | 82.54 | 81.98 | |
| MemVRBase Model=LLaVA-1.52026.05 | 81.97 | 81.33 | |
| FoxBackbone=LLaVA-1.52026.06 | 81.93 | — | |
| OPERABackbone=InstructBLIP2026.06 | 81.57 | — | |
| CausalMMBackbone=InstructBLIP2026.06 | 81.32 | — | |
| CausalMMBackbone=LLaVA-1.52026.06 | 81.27 | — | |
| OPERABackbone=LLaVA-1.52026.06 | 81.13 | — | |
| SIDBackbone=LLaVA-1.52026.06 | 81.1 | — | |
| ICDBase Model=LLaVA-1.52026.05 | 81.07 | 79.96 | |
| LaSCDBase Model=InstructBLIP2026.05 | 81.03 | 81.83 | |
| OPERABase Model=LLaVA-1.52026.05 | 80.88 | 80.69 | |
| FoxBackbone=InstructBLIP2026.06 | 80.6 | — | |
| OPERABackbone=Shikra2026.06 | 79.37 | — | |
| SIDBackbone=Shikra2026.06 | 79.13 | — | |
| VCDBackbone=InstructBLIP2026.06 | 78.77 | — | |
| FoxBackbone=Shikra2026.06 | 78.77 | — | |
| SIDBackbone=InstructBLIP2026.06 | 78.6 | — | |
| CausalMMBackbone=Shikra2026.06 | 78.53 | — | |
| ICDBackbone=LLaVA-1.52026.06 | 77.9 | — | |
| ICDBackbone=InstructBLIP2026.06 | 77.7 | — | |
| SIDBase Model=InstructBLIP2026.05 | 77.64 | 80.41 | |
| DeCoBase Model=LLaVA-1.52026.05 | 77.63 | 80.64 | |
| SamplingBackbone=InstructBLIP2026.06 | 77.33 | — | |
| DoLaBase Model=LLaVA-1.52026.05 | 77.32 | 76.66 | |
| VCDBase Model=LLaVA-1.52026.05 | 77.31 | 79.28 | |
| SamplingBackbone=LLaVA-1.52026.06 | 76.77 | — | |
| VCDBase Model=InstructBLIP2026.05 | 76.42 | 78.03 | |
| RegularBase Model=LLaVA-1.52026.05 | 76.03 | 76.26 | |
| ICDBackbone=Shikra2026.06 | 75.87 | — | |
| OPERABase Model=InstructBLIP2026.05 | 75.6 | 78.37 | |
| SamplingBackbone=Shikra2026.06 | 75.57 | — | |
| VCDBackbone=LLaVA-1.52026.06 | 75.33 | — | |
| ICDBase Model=InstructBLIP2026.05 | 75.28 | 76.24 | |
| DoLaBase Model=InstructBLIP2026.05 | 74.76 | 76.58 | |
| VCDBackbone=Shikra2026.06 | 73.8 | — | |
| RegularBase Model=InstructBLIP2026.05 | 72.37 | 75.42 | |
| CausalMMLVLM backbone=LLaVA-1.52026.06 | — | 81.93 | |
| CausalMMLVLM backbone=InstructBLIP2026.06 | — | 82.23 | |
| CausalMMLVLM backbone=Shikra2026.06 | — | 79.47 | |
| FoxLVLM backbone=LLaVA-1.52026.06 | — | 82.26 | |
| FoxLVLM backbone=InstructBLIP2026.06 | — | 81.56 | |
| FoxLVLM backbone=Shikra2026.06 | — | 79.55 | |
| ICDLVLM backbone=LLaVA-1.52026.06 | — | 79.83 | |
| ICDLVLM backbone=InstructBLIP2026.06 | — | 79.82 | |
| ICDLVLM backbone=Shikra2026.06 | — | 77.89 | |
| OPERALVLM backbone=LLaVA-1.52026.06 | — | 81.87 | |
| OPERALVLM backbone=InstructBLIP2026.06 | — | 82.34 | |
| OPERALVLM backbone=Shikra2026.06 | — | 80.01 | |
| SamplingLVLM backbone=LLaVA-1.52026.06 | — | 79.06 | |
| SamplingLVLM backbone=InstructBLIP2026.06 | — | 78.8 | |
| SamplingLVLM backbone=Shikra2026.06 | — | 77.44 | |
| SIDLVLM backbone=LLaVA-1.52026.06 | — | 81.69 | |
| SIDLVLM backbone=InstructBLIP2026.06 | — | 80.34 | |
| SIDLVLM backbone=Shikra2026.06 | — | 80.2 | |
| VCDLVLM backbone=LLaVA-1.52026.06 | — | 78 | |
| VCDLVLM backbone=InstructBLIP2026.06 | — | 79.87 | |
| VCDLVLM backbone=Shikra2026.06 | — | 75.9 |