Hallucination Detection on XSum (AUROC)
0.73AUROCINTRYGUEmin-max
Evaluation Results
| Method | Links | |
|---|---|---|
| INTRYGUEmin-maxBackbone=Mistral-7B2026.03 | 0.73 | |
| INTRYGUEmeanBackbone=Mistral-7B2026.03 | 0.7 | |
| LN-EntropyBackbone=Mistral-7B2026.03 | 0.69 | |
| ReDeEPBackbone=Qwen3-8B2026.03 | 0.69 | |
| INTRYGUEmin-maxBackbone=Qwen3-8B2026.03 | 0.68 | |
| MaxEntropyBackbone=Mistral-7B2026.03 | 0.67 | |
| Semantic EntropyBackbone=Qwen3-8B2026.03 | 0.67 | |
| INTRYGUEmin-maxBackbone=Llama-2-7B2026.03 | 0.66 | |
| INTRYGUEmin-maxBackbone=Llama-3.1-8B2026.03 | 0.66 | |
| EigenScoreBackbone=Llama-3.1-8B2026.03 | 0.66 | |
| INTRYGUEmin-maxBackbone=Llama-2-13B2026.03 | 0.63 | |
| INTRYGUEmin-maxBackbone=gemma-3-4b2026.03 | 0.62 |