Spatial-relation hallucination detection on R-Bench Instance
77.39AccuracyRVE
Evaluation results
| Method | Links | |
|---|---|---|
| RVEBackbone=InstructBLIP-7B2026.05 | 77.39 | |
| RVEBackbone=ShareGPT4V-7B2026.05 | 77.3 | |
| RVEBackbone=LLaVA-NeXT-7B2026.05 | 76.29 | |
| RVEBackbone=LLaVA-1.5-7B2026.05 | 75.19 | |
| InstructBLIP-7BModel Version=Vanilla2026.05 | 69.12 | |
| ShareGPT4V-7BModel Version=Vanilla2026.05 | 64.61 | |
| LLaVA-NeXT-7BModel Version=Vanilla2026.05 | 63.51 | |
| LLaVA-1.5-7BModel Version=Vanilla2026.05 | 60.85 |