Reasoning on Natural
70.96AccuracyHALLUGUARD
Evaluation results
| Method | Links | |
|---|---|---|
| HALLUGUARDBackbone=Llama3.1-8B2026.01 | 70.96 | |
| EnergyBackbone=Llama3.1-8B2026.01 | 68.59 | |
| MINDBackbone=Llama3.1-8B2026.01 | 68.32 | |
| P(true)Backbone=Llama3.1-8B2026.01 | 68.16 | |
| Semantic Ent.Backbone=Llama3.1-8B2026.01 | 68.1 | |
| LN EntropyBackbone=Llama3.1-8B2026.01 | 68.04 | |
| FActScoreBackbone=Llama3.1-8B2026.01 | 67.74 | |
| PerplexityBackbone=Llama3.1-8B2026.01 | 67.51 | |
| InsideBackbone=Llama3.1-8B2026.01 | 67.42 | |
| RACEBackbone=Llama3.1-8B2026.01 | 66.9 | |
| SelfCheck GPTBackbone=Llama3.1-8B2026.01 | 65.68 | |
| IO PromptBackbone=Llama3.1-8B2026.01 | 55.24 |