Hallucination Detection on SelfAware
0.587AUROCNon-Contradiction
Evaluation Results
| Method | Links | |
|---|---|---|
| Non-ContradictionModel=Llama-3-8B, Scorer Family=BB2026.06 | 0.587 | |
| GPT-4oModel=Llama-3-8B, Scorer Family=Judge2026.06 | 0.578 | |
| SonnetModel=GPT-4o, Scorer Family=Judge2026.06 | 0.528 | |
| SonnetModel=Gemini-2.5, Scorer Family=Judge2026.06 | 0.528 | |
| P(True)Model=Gemini-2.5, Scorer Family=WB2026.06 | 0.51 | |
| Sequence ProbabilityModel=Llama-3-8B, Scorer Family=WB2026.06 | 0.502 | |
| Exact MatchModel=GPT-4o, Scorer Family=BB2026.06 | 0.499 | |
| Exact MatchModel=Gemini-2.5, Scorer Family=BB2026.06 | 0.499 | |
| Probability MarginModel=GPT-4o, Scorer Family=WB2026.06 | 0.483 |