Competition Mathematics on OlympiadBench (Accuracy (avg@4))
61.1Accuracy (avg@4)PAPO
Evaluation Results
| Method | Links | |
|---|---|---|
| PAPOModel=Qwen3-4B-Base2026.03 | 61.1 | |
| PAPOModel=Qwen2.5-14B2026.03 | 59.8 | |
| ORM(DAPO)Model=Qwen3-4B-Base2026.03 | 55 | |
| ORM(GRPO)Model=Qwen2.5-14B2026.03 | 54.3 | |
| PAPOModel=Qwen2.5-7B2026.03 | 51.3 | |
| ORM(GRPO)Model=Qwen2.5-7B2026.03 | 46.3 | |
| PAPOModel=Qwen2.5-3B2026.03 | 38.6 | |
| BaseModel=Qwen3-4B-Base2026.03 | 36.8 | |
| BaseModel=Qwen2.5-14B2026.03 | 35.5 | |
| ORM(GRPO)Model=Qwen2.5-3B2026.03 | 35.3 | |
| BaseModel=Qwen2.5-7B2026.03 | 32.8 | |
| BaseModel=Qwen2.5-3B2026.03 | 22.4 |