Math Reasoning on AIME 24 (Avg@32, Pass@16)
39.06Avg@32 Scoretraining-time reweighting
Evaluation results
| Method | Links | ||
|---|---|---|---|
| training-time reweightingModel=Qwen2.5-Math-7B2026.03 | 39.06 | 60.58 | |
| training-time reweightingModel=Qwen3-8B-Base2026.03 | 38.13 | 69.87 | |
| DAPOModel=Qwen3-8B-Base2026.03 | 36.98 | 72.3 | |
| DAPOModel=Qwen2.5-Math-7B2026.03 | 35.73 | 54.09 | |
| BaseModel=Qwen2.5-Math-7B2026.03 | 14.79 | 47.46 | |
| BaseModel=Qwen3-8B-Base2026.03 | 5.42 | 30.63 |