Factual consistency evaluation on CELLS entity swap perturbation
69.3Kendall's τGPT-4o
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| GPT-4o2025.03 | 69.3 | 74.3 | 98.4 | |
| PlainQAFact2025.03 | 56.9 | 68.2 | 90.2 | |
| QAFactEval2025.03 | 55.9 | 64.6 | 89.4 | |
| AlignScore2025.03 | 36.7 | 44.6 | 75.9 | |
| Llama 3.12025.03 | 24.6 | 29.9 | 65.7 | |
| QuestEval2025.03 | 22.6 | 27 | 66 | |
| DAE2025.03 | 20.6 | 20.6 | 56.8 | |
| SummaC2025.03 | 16.4 | 24.5 | 61.7 |