Hallucination Detection on Pool (test)
0.8598AuROCspilled energy
Evaluation results
| Method | Links | |
|---|---|---|
| spilled energyLLM=LLaMA-Instruct, Aggregation Strategy=Min, Notation=Spilled ΔE2026.02 | 0.8598 | |
| spilled energyLLM=LLaMA, Aggregation Strategy=Min, Notation=Spilled ΔE2026.02 | 0.7904 | |
| marginalized energyLLM=LLaMA, Aggregation Strategy=Max, Notation=Marginal E^m2026.02 | 0.78 | |
| spilled energyLLM=LLaMA, Aggregation Strategy=Min, Notation=Spilled ΔEs2026.02 | 0.7775 | |
| marginalized energyLLM=LLaMA-Instruct, Aggregation Strategy=Max, Notation=Marginal E^m2026.02 | 0.7672 | |
| marginalized energyLLM=LLaMA-Instruct, Aggregation Strategy=Min, Notation=Marginal E^m2026.02 | 0.7591 | |
| Logit E^lLLM=LLaMA-Instruct, Aggregation Strategy=Max2026.02 | 0.7285 | |
| Orgad et al.LLM=LLaMA-Instruct, Aggregation Strategy=Mean2026.02 | 0.6656 | |
| Orgad et al.LLM=LLaMA, Aggregation Strategy=Mean2026.02 | 0.6122 | |
| marginalized energyLLM=LLaMA, Aggregation Strategy=Min, Notation=Marginal E^m2026.02 | 0.5839 | |
| spilled energyLLM=LLaMA-Instruct, Aggregation Strategy=Max, Notation=Spilled ΔEs2026.02 | 0.5365 | |
| Logit E^lLLM=LLaMA, Aggregation Strategy=Max2026.02 | 0.5347 |