Discriminative Object Hallucination on AMBER Discriminative Task
87.4F1 ScoreGPT-4V
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GPT-4VBase Model=Proprietary2026.05 | 87.4 | 83.4 | |
| HALVABase Model=LLaVA-v1.5-13B2026.05 | 86.5 | — | |
| HSA-DPOBase Model=LLaVA-v1.5-13B2026.05 | 86.1 | 80.8 | |
| LBPBase Model=LLaVA-v1.5-7B2026.05 | 86.1 | 78.6 | |
| LBPBase Model=LLaVA-v1.5-13B2026.05 | 85.4 | 77 | |
| AMP-MEGBase Model=LLaVA-v1.5-13B2026.05 | 84.6 | 79.5 | |
| RLAIF-VBase Model=LLaVA-v1.5-7B2026.05 | 84.5 | 76.8 | |
| LLaVA-v1.5-13BBase Model=LLaVA-v1.5-13B2026.05 | 84.4 | 79.3 | |
| OctopusVLM=LLaVA-1.52026.04 | 82.7 | 76.7 | |
| IECD2VLM=LLaVA-1.52026.04 | 82.6 | 77.8 | |
| IECD2VLM=InstructBLIP2026.04 | 82.1 | 76.6 | |
| CCA-LLaVABase Model=LLaVA-v1.5-7B2026.05 | 81.9 | 77.7 | |
| V-DPOBase Model=LLaVA-v1.5-7B2026.05 | 81.6 | — | |
| OctopusVLM=InstructBLIP2026.04 | 79.7 | 74 | |
| ICDVLM=InstructBLIP2026.04 | 75.9 | 69.6 | |
| VCDVLM=InstructBLIP2026.04 | 75.6 | 66.3 | |
| RLHF-VBase Model=LLaVA-v1.5-13B2026.05 | 75 | 72.6 | |
| LLaVA-v1.5-7BBase Model=LLaVA-v1.5-7B2026.05 | 74.7 | 72 | |
| RegularVLM=InstructBLIP2026.04 | 74.6 | 68.2 | |
| ICDVLM=LLaVA-1.52026.04 | 72.8 | 69.9 | |
| RegularVLM=LLaVA-1.52026.04 | 72 | 69.2 | |
| VCDVLM=LLaVA-1.52026.04 | 71.1 | 67.3 |