Answer-level hallucination detection on RAGTruth
75.96APCORTEX
Evaluation results
| Method | Links | ||
|---|---|---|---|
| CORTEXBackbone LLM=Qwen2026.06 | 75.96 | 90.77 | |
| RAGLensBackbone LLM=Llama2026.06 | 73.29 | 90.11 | |
| CORTEXBackbone LLM=Llama2026.06 | 68.93 | 89.57 | |
| CORTEXBackbone LLM=Mistral2026.06 | 67.87 | 89.07 | |
| RAGLensBackbone LLM=Mistral2026.06 | 67.81 | 88.19 | |
| SAPLMABackbone LLM=Llama2026.06 | 67.08 | 88.63 | |
| RAGLensBackbone LLM=Qwen2026.06 | 66.79 | 88.12 | |
| SAPLMABackbone LLM=Qwen2026.06 | 59.01 | 86 | |
| SAPLMABackbone LLM=Mistral2026.06 | 48.97 | 80.77 | |
| LLM-CheckBackbone LLM=Llama2026.06 | 33.75 | 71 | |
| LLM-CheckBackbone LLM=Qwen2026.06 | 33.3 | 71.98 | |
| LLM-CheckBackbone LLM=Mistral2026.06 | 32.4 | 71.7 | |
| NLLBackbone LLM=Mistral2026.06 | 23.13 | 60.88 | |
| ICR ProbeBackbone LLM=Qwen2026.06 | 23.01 | 61.62 | |
| NLLBackbone LLM=Qwen2026.06 | 20.19 | 52.27 | |
| NLLBackbone LLM=Llama2026.06 | 19.94 | 56.21 | |
| ICR ProbeBackbone LLM=Mistral2026.06 | 18.31 | 51.82 | |
| ICR ProbeBackbone LLM=Llama2026.06 | 12.86 | 34.59 |