Hallucination Detection (Dialogue) on HaluEval DA
77.1F1 ScoreHalluClean
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| HalluCleanBackbone=DeepSeek-V32025.11 | 77.1 | 72.5 | |
| GPT-4o-miniStrategy=Direct Ask2025.11 | 76.5 | 74 | |
| DeepSeek-V3Strategy=Direct Ask2025.11 | 75.4 | 71.8 | |
| HalluCleanBackbone=GPT-3.5-turbo2025.11 | 74.3 | 69.3 | |
| HalluCleanBackbone=Llama-3-70B2025.11 | 74 | 68.8 | |
| ChatProtectBackbone=GPT-3.5-turbo2025.11 | 72 | 69.3 | |
| DeepSeek-R1Strategy=Direct Ask2025.11 | 71 | 69.8 | |
| Llama-3-70BStrategy=Direct Ask2025.11 | 66.4 | 62.3 | |
| GPT-3.5-turboStrategy=Direct Ask2025.11 | 62.8 | 66 | |
| Step-by-StepBackbone=GPT-3.5-turbo2025.11 | 61.4 | 65.8 | |
| SelfCheckGPTBackbone=GPT-3.5-turbo2025.11 | 19.9 | 27.8 | |
| Plan-and-SolveBackbone=GPT-3.5-turbo2025.11 | 19.3 | 52 |