Mathematical Reasoning on Minerva (Avg@8, Pass@8)
40.81Avg@8GRPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GRPOBase Model=Qwen3-4B-Base, Training Data=DAPO-Math-14k2026.03 | 40.81 | 55.51 | |
| OPDBase Model=Qwen3-4B-Base, Training Data=DAPO-Math-14k2026.03 | 40.08 | 54 | |
| EOPDBase Model=Qwen3-4B-Base, Training Data=DAPO-Math-14k2026.03 | 39.71 | 56.99 | |
| KDBase Model=Qwen3-4B-Base, Training Data=DAPO-Math-14k2026.03 | 34.65 | 55.15 | |
| EOPDBase Model=Qwen3-1.7B-Base, Training Data=MATH2026.03 | 30.15 | 50.74 | |
| OPDBase Model=Qwen3-1.7B-Base, Training Data=MATH2026.03 | 29.83 | 47.06 | |
| GRPOBase Model=Qwen3-1.7B-Base, Training Data=MATH2026.03 | 29.46 | 48.16 | |
| KDBase Model=Qwen3-1.7B-Base, Training Data=MATH2026.03 | 28.14 | 44.1 | |
| EOPDBase Model=Qwen3-0.6B-Base, Training Data=MATH2026.03 | 20.82 | 37.13 | |
| GRPOBase Model=Qwen3-0.6B-Base, Training Data=MATH2026.03 | 20.08 | 35.66 | |
| OPDBase Model=Qwen3-0.6B-Base, Training Data=MATH2026.03 | 19.98 | 34.93 | |
| KDBase Model=Qwen3-0.6B-Base, Training Data=MATH2026.03 | 18.61 | 36.03 |