Hallucination Detection on Generated responses (test)
92.5AccuracyFisher-based label propagation
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Fisher-based label propagationModel=yi1.5-9B2026.02 | 92.5 | 95.3 | 0.5 | -1.1 | 0.9 | 1.1 | |
| Fisher-based label propagationModel=deepseek-7B2026.02 | 90.9 | 94.1 | 0.4 | -1.1 | 0.8 | 1.2 | |
| Fisher-based label propagationModel=llama3-8B2026.02 | 89.1 | 92.5 | 0.3 | -0.9 | 0.7 | 1 | |
| Fisher-based label propagationModel=gemma2-27B2026.02 | 88.3 | 93 | 0 | -0.8 | 0.6 | 0.8 | |
| Fisher-based label propagationModel=mistral-7B2026.02 | 86.8 | 92.1 | -0.1 | -0.6 | 0.4 | 0.7 | |
| Fisher-based label propagationModel=gemma2-9B2026.02 | 86.1 | 91.5 | 0 | -0.7 | 0.5 | 0.7 | |
| Fisher-based label propagationModel=qwen2.5-32B2026.02 | 85.5 | 89.1 | 0.2 | -0.6 | 0.4 | 0.6 | |
| Fisher-based label propagationModel=phi4-14B2026.02 | 85.2 | 84 | 0.3 | -0.4 | 0.4 | 0.5 | |
| Fisher-based label propagationModel=qwen2.5-14B2026.02 | 85 | 88.7 | 0.2 | -0.5 | 0.4 | 0.6 | |
| Fisher-based label propagationModel=solar1-11B2026.02 | 82.7 | 86.9 | 0.2 | -0.5 | 0.5 | 0.6 |