Hallucination Detection on Hellaswag
0.696AUROCFunction Vector
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Function VectorModel=LLaMA2-13B2026.04 | 0.696 | — | |
| UQ_ICLModel=LLaMA2-13B2026.04 | 0.685 | — | |
| Self-FVModel=LLaMA2-13B2026.04 | 0.685 | — | |
| Total EntropyModel=LLaMA2-13B2026.04 | 0.673 | — | |
| Semantic EntropyModel=LLaMA2-13B2026.04 | 0.665 | — | |
| Function VectorModel=LLaMA2-7B2026.04 | 0.624 | — | |
| MaxProbModel=LLaMA2-7B2026.04 | 0.624 | — | |
| Self-FVModel=LLaMA2-7B2026.04 | 0.619 | — | |
| Total EntropyModel=LLaMA2-7B2026.04 | 0.608 | — | |
| Semantic EntropyModel=LLaMA2-7B2026.04 | 0.578 | — | |
| UQ_ICLModel=LLaMA2-7B2026.04 | 0.575 | — | |
| Lookback LensModel=LLaMA2-7B2026.04 | 0.53 | — | |
| Function VectorModel=LLaMA2-7B2026.04 | — | 29.1 | |
| Function VectorModel=LLaMA2-13B2026.04 | — | 46.9 | |
| Self-FVModel=LLaMA2-7B2026.04 | — | 27.3 | |
| Self-FVModel=LLaMA2-13B2026.04 | — | 45.8 | |
| Semantic EntropyModel=LLaMA2-7B2026.04 | — | 19.1 | |
| Semantic EntropyModel=LLaMA2-13B2026.04 | — | 39.4 | |
| Total EntropyModel=LLaMA2-7B2026.04 | — | 24.2 | |
| Total EntropyModel=LLaMA2-13B2026.04 | — | 43 | |
| UQ_ICLModel=LLaMA2-7B2026.04 | — | 9.6 | |
| UQ_ICLModel=LLaMA2-13B2026.04 | — | 22.9 |