Hallucination Evaluation on MMHal-Bench-V
2.57Hallucination ScoreSENTINEL
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SENTINELBackbone=LLaVA-1.5-13B2026.04 | 2.57 | 49 | |
| LLaVA-1.5-13BBackbone=LLaVA-1.5-13B2026.04 | 2.47 | 52 | |
| AVES-DPOBackbone=LLaVA-1.5-13B2026.04 | 2.41 | 51 | |
| SENTINELBackbone=LLaVA-1.5-7B2026.04 | 2.38 | 54 | |
| AVES-DPOBackbone=LLaVA-1.5-7B2026.04 | 2.32 | 56 | |
| POVIDBackbone=LLaVA-1.5-7B2026.04 | 2.3 | 57 | |
| LLaVA-1.5-7BBackbone=LLaVA-1.5-7B2026.04 | 2.22 | 58 | |
| LLaVA-RLHFBackbone=LLaVA-1.5-7B2026.04 | 1.55 | 78 | |
| LLaVA-RLHFBackbone=LLaVA-1.5-13B2026.04 | 1.35 | 86 |