Mathematical Problem Solving on AIME 2025 (Pass@k Series)
12.9Pass@1CaSP
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CaSPBackbone=Qwen2.5-Math-7B2025.10 | 12.9 | 19.2 | 25.4 | 30.9 | 35.8 | 40.4 | 44.7 | 49.5 | 55.6 | 62.5 | 68.9 | |
| GRPOBackbone=Qwen2.5-Math-7B2025.10 | 12 | 17.3 | 23.4 | 29.5 | 34.7 | 39.2 | 43.5 | 47.9 | 53.1 | 58.3 | 62.3 | |
| Base ModelBackbone=Qwen2.5-Math-7B2025.10 | 5.6 | 9.5 | 14.7 | 20.8 | 27 | 32.9 | 38.6 | 44.3 | 50.5 | 57.9 | 65.5 |