Token-level Hallucination Detection on AIME 2025
6.46S_incor
Qwen3-0.6B
Updated 21d ago
Evaluation Results
| Method | Links | ||
|---|---|---|---|
2026.05 | 6.46 | 95.84 | |
2026.05 | 18.76 | 100 | |
2026.05 | 40.14 | 100 | |
2026.05 | 44.75 | 100 | |
2026.05 | 48.92 | 99.72 | |
2026.05 | 56.18 | 99.31 | |
2026.05 | 56.24 | 100 | |
2026.05 | 64.34 | 89.31 | |
2026.05 | 67.02 | 96.41 | |
2026.05 | 67.32 | 92.66 | |
2026.05 | 68.88 | 95.84 | |
2026.05 | 76.09 | 100 | |
2026.05 | 85.88 | 100 | |
2026.05 | 87.89 | 100 |