Hallucination Evaluation on POPE COCO 2014 (Random)
90.1AccuracyVEGAS
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VEGASBase Model=LLaVA-1.5, Decoding Strategy=Greedy2025.12 | 90.1 | 89.98 | |
| PAIBase Model=LLaVA-1.5, Decoding Strategy=Greedy2025.12 | 90.03 | 89.98 | |
| VISTABase Model=LLaVA-1.5, Decoding Strategy=Greedy2025.12 | 90.03 | 90.02 | |
| VanillaBase Model=LLaVA-1.5, Decoding Strategy=Greedy2025.12 | 89.33 | 89.29 | |
| VCDBase Model=LLaVA-1.5, Decoding Strategy=Greedy2025.12 | 89.05 | 89.03 | |
| VEGASBase Model=Shikra, Decoding Strategy=Greedy2025.12 | 84.73 | 84.17 | |
| VISTABase Model=Shikra, Decoding Strategy=Greedy2025.12 | 84.38 | 84.01 | |
| VEGASBase Model=MiniGPT-4, Decoding Strategy=Greedy2025.12 | 84.07 | 81.58 | |
| VISTABase Model=MiniGPT-4, Decoding Strategy=Greedy2025.12 | 83.5 | 81.43 | |
| VanillaBase Model=Shikra, Decoding Strategy=Greedy2025.12 | 83.36 | 83.52 | |
| PAIBase Model=Shikra, Decoding Strategy=Greedy2025.12 | 83.3 | 83.53 | |
| VanillaBase Model=MiniGPT-4, Decoding Strategy=Greedy2025.12 | 82.33 | 80.64 | |
| PAIBase Model=MiniGPT-4, Decoding Strategy=Greedy2025.12 | 82.3 | 80.73 |