Multilingual Multimodal Multiple-Choice Question Answering on Afri-MCQA
45.2Average AccuracyDiM³
Evaluation results
| Method | Links | |||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| DiM³Backbone=Qwen32026.05 | 45.2 | 49.3 | 49.8 | 36.2 | 45 | 44.5 | 64.5 | 40.7 | 55.5 | 50.5 | 38.8 | 44.5 | 42 | 48 | 40.6 | 27.7 | 46 | |
| Qwen3-VL-8B-InstructBackbone=Qwen32026.05 | 38.3 | 37.7 | 34.2 | 34.7 | 41.5 | 33.5 | 46 | 39.2 | 40.5 | 35.7 | 43 | 42 | 35 | 38 | 30.2 | 32.3 | 49 | |
| DiM³Backbone=Qwen22026.05 | 37 | 31.7 | 34.2 | 40.7 | 42.5 | 32 | 48 | 37.6 | 34.5 | 34.7 | 38.8 | 39 | 31.5 | 39 | 33.3 | 33.1 | 41.9 | |
| Pangea-7BBackbone=Qwen22026.05 | 28.3 | 35.7 | 27.1 | 35.2 | 32 | 13 | 37 | 25.3 | 31 | 28.7 | 33.2 | 38 | 18.5 | 26.5 | 19.8 | 13.1 | 38.9 | |
| DiM3-7BModel=DiM3-7B2026.05 | 26.31 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LLaVA-v1.5-7BModel=LLaVA-v1.5-7B2026.05 | 24.61 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Palo-7BModel=Palo-7B2026.05 | 24.47 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Task ArithmeticModel=Task Arithmetic2026.05 | 24.39 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DAREModel=DARE2026.05 | 24.32 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Palo-13BModel=Palo-13B2026.05 | 23.95 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| STFModel=STF2026.05 | 23.86 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TIES-MergingModel=TIES-Merging2026.05 | 23.84 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| NeuroMergingModel=NeuroMerging2026.05 | 23.67 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BreadcrumbsModel=Breadcrumbs2026.05 | 21 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PCB-MergingModel=PCB-Merging2026.05 | 0 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — |