Mathematical Reasoning on AIME 2025 (Pass@1 to Pass@256)
1.18Pass@1CW-NSR
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| CW-NSRBackbone=Qwen2.5-Math-1.5B2026.05 | 1.18 | 2.19 | 3.82 | 6.01 | 8.37 | 10.48 | 12.21 | 14.16 | 16.67 | |
| A-NSRBackbone=Qwen2.5-Math-1.5B2026.05 | 0.96 | 1.79 | 3.16 | 5.09 | 7.35 | 9.71 | 11.7 | 12.91 | 13.33 | |
| W-REINFORCEBackbone=Qwen2.5-Math-1.5B2026.05 | 0.8 | 1.52 | 2.74 | 4.57 | 6.87 | 9.47 | 12.2 | 14.56 | 16.67 |