Hallucination Tracing on Mistral
78.95Recall@kHART
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| HARTk=52026.03 | 78.95 | 77.39 | 59.21 | 76.86 | |
| HARTk=22026.03 | 77.8 | 76.85 | 58.23 | 76.51 | |
| HARTk=12026.03 | 75.22 | 74.13 | 53.52 | 75.22 | |
| BM25k=52026.03 | 43.45 | 50.14 | 21.61 | 37.3 | |
| Cross-Encoderk=52026.03 | 41.36 | 48.33 | 22.51 | 35.15 | |
| Sentence-BERTk=52026.03 | 30.41 | 36.98 | 13.77 | 21.44 | |
| BM25k=22026.03 | 21.58 | 42.59 | 1.18 | 33.82 | |
| Cross-Encoderk=22026.03 | 21.51 | 43.88 | 0.76 | 32.14 | |
| Sentence-BERTk=22026.03 | 15.91 | 32.52 | 0.42 | 17.94 | |
| DPRk=52026.03 | 11.28 | 17.49 | 1.69 | 16.99 | |
| Cross-Encoderk=12026.03 | 10.42 | 30.85 | 0.31 | 26.63 | |
| BM25k=12026.03 | 9.58 | 28.88 | 0.25 | 28.82 | |
| Sentence-BERTk=12026.03 | 8.11 | 24.02 | 0.17 | 13.79 | |
| DPRk=22026.03 | 6.07 | 13.68 | 0.2 | 15.3 | |
| DPRk=12026.03 | 3.3 | 9.58 | 0.11 | 12.97 |