Multiple Choice Question Answering on ArabCulture native prompts (test)
47.9AccuracyQwen2.5-7B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen2.5-7BAugmentation=None, Prompting=Base2026.01 | 47.9 | |
| Qwen2.5-7B-ITAugmentation=None, Prompting=Base2026.01 | 39.5 | |
| Qwen2.5-3BAugmentation=N-Asrt, Prompting=CCKG Native Assertions2026.01 | 39.3 | |
| Qwen2.5-1.5BAugmentation=None, Prompting=Base2026.01 | 38.8 | |
| Gemma2-9BAugmentation=N-Asrt, Prompting=CCKG Native Assertions2026.01 | 38.1 | |
| Qwen2.5-7B-ITAugmentation=E-Path, Prompting=CCKG English Paths2026.01 | 38 | |
| Qwen2.5-0.5BAugmentation=N-Path, Prompting=CCKG Native Paths2026.01 | 34.5 | |
| Llama3.2-3BAugmentation=Mango, Prompting=CCKG2026.01 | 34.4 | |
| Llama3.1-8B-ITAugmentation=E-Path, Prompting=CCKG English Paths2026.01 | 34.4 | |
| Llama3.2-1BAugmentation=Mango, Prompting=CCKG2026.01 | 34.2 |