Token-level Addition on Addition T3 (val)
89.56Token Accuracy
Dual-stream consistency check method
Updated 1mo ago
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| 89.56 | 30.46 | 98.13 | ||
2026.05 | 89.13 | 31.73 | 97.65 | |
2026.05 | 88.27 | 30.58 | 96.97 | |
| 87.27 | 44.39 | 94.07 | ||
2026.05 | 86.26 | - | - | |
2026.05 | 79.9 | 0.08 | 99.98 |