Discriminative Hallucination Detection on HallusionBench
73AccuracyInternVL-3.5 + FINER-Tuning
Evaluation Results
| Method | Links | |
|---|---|---|
| InternVL-3.5 + FINER-TuningSize=8B2026.03 | 73 | |
| InternVL-3.5 + FINER-TuningSize=14B2026.03 | 71.2 | |
| InternVL-3.5Size=8B2026.03 | 71 | |
| InternVL-3.5Size=14B2026.03 | 69.5 | |
| Qwen2.5-VL + FINER-TuningSize=7B2026.03 | 68.5 | |
| Qwen2.5-VLSize=7B2026.03 | 65.4 | |
| OmniLMMSize=12B2026.03 | 54.9 | |
| OmniLMM + RLAIF-VSize=12B2026.03 | 53.7 | |
| LLaVA-1.6 + FINER-TuningSize=7B2026.03 | 36.3 | |
| LLaVA-1.6Size=7B2026.03 | 33 |