Mathematical Problem Solving on AIME 2024 and 2025 (test)
38.33AccuracyCAPO
Evaluation Results
| Method | Links | |
|---|---|---|
| CAPOModel=Qwen2.5-Math-7B2026.04 | 38.33 | |
| GRPOModel=Qwen2.5-Math-7B2026.04 | 33.33 | |
| GSPOModel=Qwen2.5-Math-7B2026.04 | 32.21 | |
| CoDaPOModel=Qwen2.5-Math-7B2026.04 | 31.66 | |
| CDEModel=Qwen2.5-Math-7B2026.04 | 31.66 | |
| CAPOModel=Qwen2.5-Math-1.5B2026.04 | 25.33 | |
| SimKOModel=Qwen2.5-Math-7B2026.04 | 23.33 | |
| CoDaPOModel=Qwen2.5-Math-1.5B2026.04 | 21.67 | |
| GRPOModel=Qwen2.5-Math-1.5B2026.04 | 20.33 | |
| GSPOModel=Qwen2.5-Math-1.5B2026.04 | 20 | |
| CDEModel=Qwen2.5-Math-1.5B2026.04 | 16.67 | |
| SimKOModel=Qwen2.5-Math-1.5B2026.04 | 11.67 |