Hallucination Detection on AG News (AUROC)
0.852AUROCUQ_ICL
Evaluation Results
| Method | Links | |
|---|---|---|
| UQ_ICLModel=LLaMA2-13B2026.04 | 0.852 | |
| Function VectorModel=LLaMA2-7B2026.04 | 0.85 | |
| Function VectorModel=LLaMA2-13B2026.04 | 0.848 | |
| Self-FVModel=LLaMA2-7B2026.04 | 0.845 | |
| Self-FVModel=LLaMA2-13B2026.04 | 0.845 | |
| Total EntropyModel=LLaMA2-7B2026.04 | 0.843 | |
| MaxProbModel=LLaMA2-7B2026.04 | 0.837 | |
| Total EntropyModel=LLaMA2-13B2026.04 | 0.832 | |
| UQ_ICLModel=LLaMA2-7B2026.04 | 0.805 | |
| Semantic EntropyModel=LLaMA2-13B2026.04 | 0.598 | |
| Lookback LensModel=LLaMA2-7B2026.04 | 0.572 | |
| Semantic EntropyModel=LLaMA2-7B2026.04 | 0.532 |