Hallucination Detection on PsiloQA (Accuracy)
88.43AccuracyID
Evaluation Results
| Method | Links | |
|---|---|---|
| IDModel=LlaMA-3.2-3B, Forward horizon (w)=7, Representation extraction (FST)=true2026.05 | 88.43 | |
| FEPoIDModel=LlaMA-3.2-3B, Forward horizon (w)=7, Representation extraction (FST)=true2026.05 | 88.43 | |
| Val LossModel=LlaMA-3.2-3B, Forward horizon (w)=7, Representation extraction (FST)=true2026.05 | 88.22 | |
| CurvatureModel=LlaMA-3.2-3B, Forward horizon (w)=7, Representation extraction (FST)=true2026.05 | 86.87 | |
| Val LossModel=LlaMA-3.2-1B, Forward horizon (w)=3, Representation extraction (FST)=true2026.05 | 83.23 | |
| FEPoIDModel=LlaMA-3.2-1B, Forward horizon (w)=3, Representation extraction (FST)=true2026.05 | 83.23 | |
| RGNModel=LlaMA-3.2-3B, Forward horizon (w)=7, Representation extraction (FST)=true2026.05 | 82.95 | |
| CurvatureModel=LlaMA-3.2-1B, Forward horizon (w)=3, Representation extraction (FST)=true2026.05 | 80.97 | |
| SNRModel=LlaMA-3.2-1B, Forward horizon (w)=3, Representation extraction (FST)=true2026.05 | 80.97 | |
| RankMEModel=LlaMA-3.2-3B, Forward horizon (w)=7, Representation extraction (FST)=true2026.05 | 79.5 | |
| IDModel=LlaMA-3.2-1B, Forward horizon (w)=3, Representation extraction (FST)=true2026.05 | 78.97 | |
| RGNModel=LlaMA-3.2-1B, Forward horizon (w)=3, Representation extraction (FST)=true2026.05 | 77.88 | |
| SNRModel=LlaMA-3.2-3B, Forward horizon (w)=7, Representation extraction (FST)=true2026.05 | 76.65 | |
| RankMEModel=LlaMA-3.2-1B, Forward horizon (w)=3, Representation extraction (FST)=true2026.05 | 71.51 |