Multi-image and Video Understanding on MuirBench
41.4AccuracyLLaVA-OV + IRA
Evaluation Results
| Method | Links | |
|---|---|---|
| LLaVA-OV + IRABase Model=LLaVA-OV, Training Stage=IRA2026.07 | 41.4 | |
| LLaVA-OV + SFTBase Model=LLaVA-OV, Training Stage=SFT2026.07 | 40.9 | |
| InternVL2 + SFTBase Model=InternVL2, Training Stage=SFT2026.07 | 38.3 | |
| LLaVA-OV-PtBase Model=LLaVA-OV, Training Stage=Pretrain2026.07 | 37.4 | |
| InternVL2 + IRABase Model=InternVL2, Training Stage=IRA2026.07 | 36.8 | |
| InternVL2.5 + IRABase Model=InternVL2.5, Training Stage=IRA2026.07 | 35.4 | |
| InternVL2.5 + SFTBase Model=InternVL2.5, Training Stage=SFT2026.07 | 35.2 | |
| InternVL2.5-PtBase Model=InternVL2.5, Training Stage=Pretrain2026.07 | 31.6 | |
| InternVL2-PtBase Model=InternVL2, Training Stage=Pretrain2026.07 | 31.1 |