Hallucination Detection on TriviaQA Amharic - Low (test)
75.8AUROCMultiHaluDet
Evaluation Results
| Method | Links | |
|---|---|---|
| MultiHaluDetBackbone=Mistral-7B2026.05 | 75.8 | |
| MultiHaluDetBackbone=LLaMA2-7B2026.05 | 73.4 | |
| Best BaselineBackbone=Mistral-7B2026.05 | 54.1 | |
| Best BaselineBackbone=LLaMA2-7B2026.05 | 52.6 |