Hallucination Detection on RAGTruth LLaMA2-7B
0.8328RecallTPA
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| TPALLM=LLaMA2-7B2025.12 | 0.8328 | 0.7873 | 0.7238 | |
| NovoLLM=LLaMA2-7B2025.12 | 0.8274 | 0.7608 | 0.7057 | |
| ReDeEPLLM=LLaMA2-7B2025.12 | 0.8097 | 0.7458 | 0.719 | |
| ChainPollLLM=LLaMA2-7B2025.12 | 0.7832 | 0.6738 | 0.7066 | |
| SEPLLM=LLaMA2-7B2025.12 | 0.7477 | 0.7143 | 0.6627 | |
| EigenScoreLLM=LLaMA2-7B2025.12 | 0.7469 | 0.6045 | 0.6682 | |
| PromptLLM=LLaMA2-7B2025.12 | 0.72 | — | 0.672 | |
| TrulensLLM=LLaMA2-7B2025.12 | 0.6814 | 0.651 | 0.6567 | |
| RAGASLLM=LLaMA2-7B2025.12 | 0.6327 | 0.729 | 0.6667 | |
| RefCheckLLM=LLaMA2-7B2025.12 | 0.628 | 0.6912 | 0.6736 | |
| TSVLLM=LLaMA2-7B2025.12 | 0.5526 | 0.6609 | 0.6632 | |
| ITILLM=LLaMA2-7B2025.12 | 0.5416 | 0.7161 | 0.6745 | |
| LN-EntropyLLM=LLaMA2-7B2025.12 | 0.5383 | 0.5912 | 0.6655 | |
| FocusLLM=LLaMA2-7B2025.12 | 0.5309 | 0.6233 | 0.6622 | |
| P(True)LLM=LLaMA2-7B2025.12 | 0.5194 | 0.7093 | 0.5313 | |
| PerplexityLLM=LLaMA2-7B2025.12 | 0.519 | 0.5091 | 0.6749 | |
| SAPLMALLM=LLaMA2-7B2025.12 | 0.5091 | 0.7037 | 0.6726 | |
| EnergyLLM=LLaMA2-7B2025.12 | 0.5057 | 0.5619 | 0.6657 | |
| SelfCheckGPTLLM=LLaMA2-7B2025.12 | 0.4642 | — | 0.4642 |