Multilingual Vision-Language Understanding on 14 Multilingual Tasks Suite (various)
38.5Task 1 ScorePangea
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| PangeaTask Set=All 14 tasks2025.01 | 38.5 | 38.6 | 46.9 | 44.2 | 49.9 | 59.8 | |
| PangeaTask Set=Without multi-image tasks (MaRVL, VGR, VLOD)2025.01 | 38.5 | 46.5 | 47.7 | 44.4 | 49.9 | 64.9 | |
| Centurio QwenTask Set=All 14 tasks2025.01 | 38.1 | 51 | 48.3 | 47 | 50.9 | 66.6 | |
| Centurio QwenTask Set=Without multi-image tasks (MaRVL, VGR, VLOD)2025.01 | 38.1 | 49.5 | 45.6 | 45.8 | 49.6 | 66 | |
| Centurio AyaTask Set=All 14 tasks2025.01 | 35.1 | 46.4 | 47 | 46.7 | 48.3 | 60.6 | |
| Centurio AyaTask Set=Without multi-image tasks (MaRVL, VGR, VLOD)2025.01 | 35.1 | 44.5 | 45.7 | 46.2 | 47.7 | 60.7 | |
| Qwen2VL 7BTask Set=All 14 tasks2025.01 | 30.6 | 36.8 | 40.5 | 46.2 | 48 | 56.8 | |
| Qwen2VL 7BTask Set=Without multi-image tasks (MaRVL, VGR, VLOD)2025.01 | 30.6 | 38.7 | 40.8 | 46.8 | 48.3 | 61.7 | |
| InternVL 2.5 8BTask Set=All 14 tasks2025.01 | 29.9 | 37 | 37.4 | 41 | 50.5 | 64.4 | |
| InternVL 2.5 8BTask Set=Without multi-image tasks (MaRVL, VGR, VLOD)2025.01 | 29.9 | 40.4 | 35.2 | 39.4 | 49.7 | 62.3 |