Multi-choice Video Question Answering on Causal-VidQA
75.3Descriptive AccuracyMAMA-VGT
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| MAMA-VGTBackbone=VGT2026.01 | 75.3 | 77.1 | 62.2 | 68.2 | |
| LaViLa-VGTBackbone=VGT2026.01 | 74.7 | 76.1 | 61.6 | 65.7 | |
| VGT2026.01 | 74.4 | 75.6 | 60.7 | 65.6 | |
| Vid2Seq-VGTBackbone=VGT2026.01 | 73.7 | 76 | 61.3 | 65.6 | |
| MAMA-VIOLETBackbone=VIOLET2026.01 | 72.7 | 68 | 60.5 | 59.5 | |
| VIOLET2026.01 | 67.6 | 66.6 | 57.1 | 57.6 | |
| MERLOT2026.01 | 67.2 | 65.7 | 57.2 | 57 | |
| OmniVL2026.01 | 67 | 66 | 56.5 | 57.1 | |
| VindLU2026.01 | 63.6 | 54.8 | 57.7 | 54.4 | |
| LAVENDER2026.01 | 62 | 61.6 | 46.3 | 50.4 | |
| All-in-one2026.01 | 60.4 | 51.3 | 51.2 | 50.7 | |
| Singularity2026.01 | 54.3 | 49.6 | 41.3 | 46.9 |