Mathematical Problem Solving on AIME 2024 (Pass@1 to Pass@1024)
30.3Pass@1CaSP
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CaSPBackbone=Qwen2.5-Math-7B2025.10 | 30.3 | 39.4 | 46.3 | 51.9 | 57.1 | 61.6 | 66 | 71.1 | 76.5 | 81.2 | 85.6 | |
| GRPOBackbone=Qwen2.5-Math-7B2025.10 | 27 | 33.3 | 39.2 | 45 | 51.4 | 58.1 | 64.6 | 70.4 | 74.4 | 76.2 | 76.7 | |
| Base ModelBackbone=Qwen2.5-Math-7B2025.10 | 13.4 | 21.3 | 29.6 | 36.7 | 42.9 | 49.1 | 55.2 | 60.8 | 66.1 | 72 | 80.8 |