Hallucination Detection on AggreFact
69.9Balanced AccuracyStacking
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| StackingAnalyzer=Gemma-2-9B2026.05 | 69.9 | 0.768 | 5.1 | — | |
| Gemma-2-9BParams=9B2026.05 | 69.9 | — | — | 6 | |
| StackingAnalyzer=Qwen2.5-7B2026.05 | 69 | 0.756 | 6.1 | — | |
| Qwen2.5-7BParams=7B2026.05 | 69 | — | — | 9 | |
| StackingAnalyzer=Pythia-1.4B2026.05 | 68.7 | 0.751 | 6.3 | — | |
| Pythia-1.4BParams=1.4B2026.05 | 68.7 | — | — | 6 | |
| StackingAnalyzer=LLaMA-3-8B2026.05 | 68.6 | 0.751 | 6.4 | — | |
| LLaMA-3-8BParams=8B2026.05 | 68.6 | — | — | 9 | |
| StackingAnalyzer=Qwen2.5-0.5B2026.05 | 67.8 | 0.743 | 7.2 | — | |
| StackingAnalyzer=Gemma-2-2B2026.05 | 67.8 | 0.741 | 7.3 | — | |
| Qwen2.5-0.5BParams=0.5B2026.05 | 67.8 | — | — | 8 | |
| Gemma-2-2BParams=2B2026.05 | 67.8 | — | — | 8 | |
| StackingAnalyzer=LLaMA-3-3B2026.05 | 67.1 | 0.733 | 7.9 | — | |
| LLaMA-3-3BParams=3B2026.05 | 67.1 | — | — | 9 |