Discriminative Hallucination Detection on AMBER
89.4AccuracyInternVL-3.5 + FINER-Tuning
Evaluation Results
| Method | Links | |
|---|---|---|
| InternVL-3.5 + FINER-TuningSize=14B2026.03 | 89.4 | |
| InternVL-3.5 + FINER-TuningSize=8B2026.03 | 88.6 | |
| InternVL-3.5Size=8B2026.03 | 88.2 | |
| InternVL-3.5Size=14B2026.03 | 88 | |
| OmniLMM + RLAIF-VSize=12B2026.03 | 87.4 | |
| OmniLMMSize=12B2026.03 | 86.9 | |
| Qwen2.5-VL + FINER-TuningSize=7B2026.03 | 85.8 | |
| Qwen2.5-VLSize=7B2026.03 | 85.2 | |
| LLaVA-1.6 + FINER-TuningSize=7B2026.03 | 85 | |
| LLaVA-1.6Size=7B2026.03 | 78.1 |