π Overall
π Per-Language
Overall accuracy across all 2,194 questions β higher is better.
| # | Model | Accuracy (%) | Correct | Total | Unparsed |
|---|
Per-language accuracy (%) β models sorted by overall accuracy.
Benchmarking LLMs on 2,194 multiple-choice questions across 12 Ghanaian languages from the NTC licensing practice tests. All models evaluated with thinking disabled for fair comparison.
Loading benchmark data from GitHubβ¦