Hallucination Detection on RAGTruth Llama-13B
89.47RecallSelfCheckGPT
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| SelfCheckGPTEvaluator Model=GPT-3.5-Turbo2026.01 | 89.47 | 43.01 | 58.1 | |
| SIRG2026.01 | 85.51 | 78.48 | 81.84 | |
| SelfCheckGPTEvaluator Model=Llama-7B2026.01 | 75.94 | 43.66 | 55.44 | |
| LRP4RAG2026.01 | 74.58 | 77.14 | 75.86 | |
| EigenScore2026.01 | 67.15 | — | 66.37 | |
| SEP2026.01 | 65.8 | — | 71.59 | |
| PromptEvaluator Model=Llama-7B2026.01 | 56.64 | 41.02 | 47.58 | |
| PromptEvaluator Model=GPT-3.5-Turbo2026.01 | 44.36 | 47.58 | 45.91 | |
| Fine-tuningEvaluator Model=Llama-7B2026.01 | 27.92 | 62.5 | 37.62 | |
| Fine-tuningEvaluator Model=Qwen2-7B2026.01 | 25.93 | 63.55 | 35.89 |