Generative Hallucination Evaluation on AMBER
90.79ScoreACPO
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ACPOModel Scale=14B, Base Model=InternVL32026.03 | 90.79 | — | — | — | — | |
| DPO-ShiftModel Scale=14B, Base Model=InternVL32026.03 | 89.97 | — | — | — | — | |
| DPOModel Scale=14B, Base Model=InternVL32026.03 | 89.78 | — | — | — | — | |
| InstructModel Scale=14B, Base Model=InternVL32026.03 | 89.68 | — | — | — | — | |
| IPOModel Scale=14B, Base Model=InternVL32026.03 | 89.66 | — | — | — | — | |
| SimPOModel Scale=14B, Base Model=InternVL32026.03 | 89.52 | — | — | — | — | |
| β-DPOModel Scale=14B, Base Model=InternVL32026.03 | 89.45 | — | — | — | — | |
| DPO-ShiftModel Scale=8B, Base Model=InternVL32026.03 | 88.39 | — | — | — | — | |
| ACPOModel Scale=8B, Base Model=InternVL32026.03 | 88.24 | — | — | — | — | |
| IPOModel Scale=8B, Base Model=InternVL32026.03 | 87.85 | — | — | — | — | |
| DPOModel Scale=8B, Base Model=InternVL32026.03 | 87.45 | — | — | — | — | |
| InstructModel Scale=8B, Base Model=InternVL32026.03 | 86.75 | — | — | — | — | |
| SimPOModel Scale=8B, Base Model=InternVL32026.03 | 86.42 | — | — | — | — | |
| β-DPOModel Scale=8B, Base Model=InternVL32026.03 | 86.34 | — | — | — | — | |
| LLaVA-1.5-7BVariant=Original2026.04 | — | 8.2 | 48.5 | 34.9 | 3.5 | |
| LLaVA-1.5-7BVariant=Ours2026.04 | — | 6.2 | 46.2 | 24.7 | 2.4 | |
| Shikra-7BVariant=Original2026.04 | — | 12.2 | 48.9 | 49.7 | 5.6 | |
| Shikra-7BVariant=Ours2026.04 | — | 11.5 | 47.5 | 44.8 | 4.6 |