Mathematical Reasoning on AIME 2025 (Turn-based Performance Tracking)
6.7Turn 1 ScoreUntrained baseline
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Untrained baselineBase Model=Qwen3-1.7B2026.05 | 6.7 | 6.7 | 0 | |
| Bi-NACBase Model=Qwen3-1.7B2026.05 | 6.7 | 10 | 3.3 |
| Method | Links | |||
|---|---|---|---|---|
| Untrained baselineBase Model=Qwen3-1.7B2026.05 | 6.7 | 6.7 | 0 | |
| Bi-NACBase Model=Qwen3-1.7B2026.05 | 6.7 | 10 | 3.3 |