| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| HallusionBench | Accuracy76.6 | 156 | 18d ago | ||
| MSCOCO | M3ID | CHAIR_i18.2 | 122 | 1mo ago | |
| POPE MS-COCO Adversarial sampling (val) | RVCD | Accuracy85.48 | 39 | 4mo ago | |
| AMBER Discriminative | HeRA | Accuracy89 | 18 | 1mo ago | |
| AMBER Generative | CHAIRi6.3 | 18 | 1mo ago | ||
| POPE (average) | CRG | F1 Score88.93 | 14 | 2mo ago | |
| HallusionBench visual questions | GPT-4V | Accuracy65.8 | 10 | 4mo ago | |
| POPE MS-COCO Popular sampling (val) | InstructBLIP-14B | Accuracy82.77 | 6 | 4mo ago | |
| POPE MS-COCO Random sampling (val) | InstructBLIP-14B | Accuracy88.57 | 6 | 4mo ago |