Hallucination Detection on NQ Open (test)
89.4AUROCHARP
Evaluation Results
| Method | Links | |
|---|---|---|
| HARPModels=LLaMA-3.1-8B, Single=true2025.09 | 89.4 | |
| HARPModels=Qwen-2.5-7B-Instruct, Single=true2025.09 | 84 | |
| EigenScoreModels=Qwen-2.5-7B-Instruct, Single=false2025.09 | 78.9 | |
| Lexical SimilarityModels=Qwen-2.5-7B-Instruct, Single=false2025.09 | 77.8 | |
| LN-EntropyModels=Qwen-2.5-7B-Instruct, Single=false2025.09 | 77.7 | |
| Semantic EntropyModels=Qwen-2.5-7B-Instruct, Single=false2025.09 | 77.7 | |
| PerplexityModels=Qwen-2.5-7B-Instruct, Single=true2025.09 | 76.5 | |
| HaloScopeModels=LLaMA-3.1-8B, Single=true2025.09 | 62.7 | |
| Lexical SimilarityModels=LLaMA-3.1-8B, Single=false2025.09 | 60.9 | |
| HaloScopeModels=Qwen-2.5-7B-Instruct, Single=true2025.09 | 60.7 | |
| Semantic EntropyModels=LLaMA-3.1-8B, Single=false2025.09 | 60.7 | |
| EigenScoreModels=LLaMA-3.1-8B, Single=false2025.09 | 56.7 | |
| LN-EntropyModels=LLaMA-3.1-8B, Single=false2025.09 | 52.7 | |
| PerplexityModels=LLaMA-3.1-8B, Single=true2025.09 | 50.3 |