Hallucination Tracing on Qwen
83.31Recall@kHART
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| HARTk=52026.03 | 83.31 | 82.08 | 64.93 | 81.65 | |
| HARTk=22026.03 | 82.6 | 81.73 | 64.34 | 81.42 | |
| HARTk=12026.03 | 80.24 | 79.15 | 62.65 | 80.24 | |
| BM25k=52026.03 | 39.45 | 45.57 | 18.99 | 39.79 | |
| Cross-Encoderk=52026.03 | 39.1 | 42.08 | 22.71 | 41.35 | |
| Sentence-BERTk=52026.03 | 30.14 | 34.07 | 15.86 | 24.87 | |
| BM25k=22026.03 | 21.96 | 38.96 | 0.43 | 36.12 | |
| Cross-Encoderk=22026.03 | 20.46 | 36.72 | 2.38 | 38.5 | |
| Sentence-BERTk=22026.03 | 15.92 | 29.29 | 2.23 | 21.38 | |
| DPRk=52026.03 | 12.32 | 16.86 | 2.69 | 18.41 | |
| BM25k=12026.03 | 10.74 | 26.59 | 0.32 | 30.16 | |
| Cross-Encoderk=12026.03 | 9.06 | 22.69 | 0.25 | 32.98 | |
| Sentence-BERTk=12026.03 | 8.59 | 21.64 | 0.2 | 16.42 | |
| DPRk=22026.03 | 6.52 | 12.9 | 0.42 | 16.35 | |
| DPRk=12026.03 | 3.49 | 8.69 | 0.14 | 13.53 |