Mathematical Reasoning on Average (AIME23, AMC23, MATH500)
75.2Majority Score (Avg)PAEC
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| PAECModel=Qwen3-4B-Base2026.06 | 75.2 | 62 | |
| GRPOModel=Qwen3-4B-Base2026.06 | 71.9 | 60.4 | |
| Global Entropy RegModel=Qwen3-4B-Base2026.06 | 69.7 | 59.5 | |
| BaseModel=Qwen3-4B-Base2026.06 | 66.7 | 57.6 | |
| PAECModel=Llama3.1-8B2026.06 | 10.9 | 6.7 | |
| GRPOModel=Llama3.1-8B2026.06 | 7.9 | 5.8 | |
| Global Entropy RegModel=Llama3.1-8B2026.06 | 7.4 | 5.5 | |
| BaseModel=Llama3.1-8B2026.06 | 3.6 | 1.6 |