Hallucination Detection on RAGTruth LLaMA3-8B
78.6RecallTPA
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| TPALLM=LLaMA3-8B2025.12 | 78.6 | 0.8211 | 78.43 | |
| ReDeEPLLM=LLaMA3-8B2025.12 | 78.19 | 0.7285 | 69.47 | |
| NovoLLM=LLaMA3-8B2025.12 | 77.37 | 0.8258 | 78.01 | |
| SEPLLM=LLaMA3-8B2025.12 | 73.33 | 0.7004 | 69.15 | |
| P(True)LLM=LLaMA3-8B2025.12 | 70.83 | 0.6323 | 68.35 | |
| EigenScoreLLM=LLaMA3-8B2025.12 | 70.78 | 0.6497 | 67.45 | |
| ITILLM=LLaMA3-8B2025.12 | 68.5 | 0.6534 | 69.33 | |
| FocusLLM=LLaMA3-8B2025.12 | 66.88 | 0.6378 | 68.79 | |
| PerplexityLLM=LLaMA3-8B2025.12 | 65.37 | 0.6235 | 67.78 | |
| LN-EntropyLLM=LLaMA3-8B2025.12 | 55.96 | 0.7021 | 62.82 | |
| TSVLLM=LLaMA3-8B2025.12 | 55.46 | 0.7769 | 64.42 | |
| EnergyLLM=LLaMA3-8B2025.12 | 55.14 | 0.5959 | 67.2 | |
| SAPLMALLM=LLaMA3-8B2025.12 | 54.32 | 0.7092 | 67.18 | |
| SelfCheckGPTLLM=LLaMA3-8B2025.12 | 51.11 | — | 51.11 | |
| ChainPollLLM=LLaMA3-8B2025.12 | 44.86 | 0.6687 | 58.13 | |
| PromptLLM=LLaMA3-8B2025.12 | 44.03 | — | 56.91 | |
| RAGASLLM=LLaMA3-8B2025.12 | 39.09 | 0.6776 | 50.94 | |
| TrulensLLM=LLaMA3-8B2025.12 | 39.09 | 0.6464 | 50.53 | |
| RefCheckLLM=LLaMA3-8B2025.12 | 35.8 | 0.6014 | 46.28 |