Hallucination Detection on HaluEval Amharic - Low (test)
78.5AUROCMultiHaluDet
Evaluation Results
| Method | Links | |
|---|---|---|
| MultiHaluDetBackbone=Mistral-7B2026.05 | 78.5 | |
| MultiHaluDetBackbone=LLaMA2-7B2026.05 | 76.2 | |
| Best BaselineBackbone=Mistral-7B2026.05 | 62.3 | |
| Best BaselineBackbone=LLaMA2-7B2026.05 | 59.8 |