Multiple-choice tasks on FACTOR
65.83Accuracy (news)TruthX
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| TruthXBackbone=Llama-2-7B-Chat2024.02 | 65.83 | 65.25 | 57.18 | |
| Llama-2-7B-ChatBackbone=Llama-2-7B-Chat2024.02 | 64.67 | 64.83 | 56.95 | |
| ITIBackbone=Llama-2-7B-Chat2024.02 | 53.28 | 51.69 | 43.82 |