Hallucination Detection on Glaive General Mixed
98.2RecallL26 Smoothness (single)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| L26 Smoothness (single)Model=Llama 3.1 8B, Configuration Type=single2026.02 | 98.2 | — | 23.6 | |
| L0 Fied.+L21 Ent.+L1 Fied. (triplet)Model=Llama 3.1 8B, Configuration Type=triplet2026.02 | 95.4 | — | 35.1 | |
| L3 Entropy (single)Model=Mistral 7B v0.1, Configuration Type=single2026.02 | 94.7 | — | 22 | |
| L0 Fied.+L1 Fied. (pair)Model=Llama 3.1 8B, Configuration Type=pair2026.02 | 94.5 | — | 35.4 | |
| L1 Fied.+L5 Sm.+L10 Fied.+L6 Fied.+L25 HFERModel=Mistral 7B v0.12026.02 | 91.3 | — | 38.3 | |
| L1 Fied.+L5 Sm.+L10 Fied.+L25 HFER (4 feat.)Model=Mistral 7B v0.1, Number of Features=42026.02 | 90.8 | — | 38.3 | |
| L19 Sm.+L17 Sm.+L17 HFER (triplet)Model=Qwen 2.5 0.5B, Configuration Type=triplet2026.02 | 86.5 | — | 45.1 | |
| L20 Sm.+L23 Sm.+L21 Ent.+L18 HFER+L17 HFERModel=Qwen 2.5 0.5B2026.02 | 86.5 | — | 40.7 | |
| L7 Smoothness (single)Model=Qwen 2.5 0.5B, Configuration Type=single2026.02 | 78 | — | 40 | |
| Semantic Similarity2026.02 | 73.1 | — | 100 |