Hallucination Detection on Glaive Finance
95.8RecallL27 Entropy (single)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| L27 Entropy (single)Model=Llama 3.1 8B, Configuration Type=single2026.02 | 95.8 | — | 64.4 | |
| L28 Fiedler+L30 Fiedler (pair)Model=Llama 3.1 8B, Configuration Type=pair2026.02 | 89.7 | — | 69.4 | |
| L29 Fied.+L30 Sm.+L28 Fied.+L30 Fied.+L19 HFERModel=Llama 3.1 8B2026.02 | 88.7 | — | 79.6 | |
| L30 Fied.+L1 Ent.+L25 HFER (triplet)Model=Llama 3.1 8B, Configuration Type=triplet2026.02 | 86.1 | — | 81 | |
| Semantic Similarity2026.02 | 73.1 | — | 100 |