Hallucination Detection on TruthfulQA EN
88.34F1-ScoreCrossHallu
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CrossHalluBackbone=Aya, Train dataset=TruthfulQA EN, Transfer setting group=Same-language baseline (BS)2026.07 | 88.34 | 82.82 | 88.85 | |
| CrossHalluBackbone=Allam, Train dataset=TruthfulQA EN, Transfer setting group=Same-language baseline (BS)2026.07 | 84.77 | 71.3 | 83.47 | |
| CrossHalluBackbone=Allam, Train dataset=TruthfulQA AR, Transfer setting group=Cross-lingual (CL)2026.07 | 83.38 | 58.19 | 76.52 | |
| CrossHalluBackbone=Mistral, Train dataset=HalluScore, Transfer setting group=Cross-lingual + cross-domain (CL-CD)2026.07 | 82.67 | 62.67 | 79.64 | |
| CrossHalluBackbone=Allam, Train dataset=HalluScore, Transfer setting group=Cross-lingual + cross-domain (CL-CD)2026.07 | 81.88 | 61.44 | 78.05 | |
| CrossHalluBackbone=Aya, Train dataset=HalluScore, Transfer setting group=Cross-lingual + cross-domain (CL-CD)2026.07 | 79.38 | 62.34 | 73.45 | |
| CrossHalluBackbone=Phi4-mini, Train dataset=TruthfulQA AR, Transfer setting group=Cross-lingual (CL)2026.07 | 79 | 66.84 | 80.02 | |
| CrossHalluBackbone=Aya, Train dataset=TruthfulQA AR, Transfer setting group=Cross-lingual (CL)2026.07 | 79 | 59.39 | 76.9 | |
| CrossHalluBackbone=Phi4-mini, Train dataset=HalluScore, Transfer setting group=Cross-lingual + cross-domain (CL-CD)2026.07 | 79 | 62.84 | 77.99 | |
| CrossHalluBackbone=Phi4-mini, Train dataset=TruthfulQA EN, Transfer setting group=Same-language baseline (BS)2026.07 | 78.81 | 70.64 | 84.3 | |
| CrossHalluBackbone=Mistral, Train dataset=TruthfulQA EN, Transfer setting group=Same-language baseline (BS)2026.07 | 78.81 | 70.64 | 84.3 | |
| CrossHalluBackbone=Silma, Train dataset=HalluScore, Transfer setting group=Cross-lingual + cross-domain (CL-CD)2026.07 | 77.07 | 59.16 | 73.08 | |
| CrossHalluBackbone=Silma, Train dataset=TruthfulQA AR, Transfer setting group=Cross-lingual (CL)2026.07 | 76.77 | 49.77 | 60.64 | |
| CrossHalluBackbone=Silma, Train dataset=TruthfulQA EN, Transfer setting group=Same-language baseline (BS)2026.07 | 76.45 | 70.01 | 75.51 | |
| CrossHalluBackbone=Mistral, Train dataset=TruthfulQA AR, Transfer setting group=Cross-lingual (CL)2026.07 | 76.19 | 66.33 | 81 | |
| CrossHalluBackbone=Qwen2.5, Train dataset=HalluScore, Transfer setting group=Cross-lingual + cross-domain (CL-CD)2026.07 | 59.22 | 58.34 | 55.1 | |
| CrossHalluBackbone=Qwen2.5, Train dataset=TruthfulQA EN, Transfer setting group=Same-language baseline (BS)2026.07 | 57.86 | 69.75 | 66.88 | |
| CrossHalluBackbone=Qwen2.5, Train dataset=TruthfulQA AR, Transfer setting group=Cross-lingual (CL)2026.07 | 6.12 | 61.56 | 55.75 |