Hallucination Detection on HaluEval French - High (test)
96.2AUROC (%)MultiHaluDet
Evaluation Results
| Method | Links | |
|---|---|---|
| MultiHaluDetBackbone=Mistral-7B2026.05 | 96.2 | |
| MultiHaluDetBackbone=LLaMA2-7B2026.05 | 95.8 | |
| Best BaselineBackbone=Mistral-7B2026.05 | 92.1 | |
| Best BaselineBackbone=LLaMA2-7B2026.05 | 89.4 |