Mathematical Reasoning on MATH 500 (Avg@8, Pass@8, Length)
64.53Avg@8PowerOPD
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| PowerOPDTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base, alpha=5002026.06 | 64.53 | 75.72 | 997 | |
| Full-vocab KL OPDTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base2026.06 | 63.12 | 87 | 4,004 | |
| PowerOPDTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base, alpha=52026.06 | 59.96 | 76.42 | 1,116 | |
| PowerOPDTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base, alpha=1002026.06 | 59.8 | 75.76 | 1,009 | |
| PowerOPDTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base, alpha=102026.06 | 59.71 | 76.1 | 1,104 | |
| PowerOPDTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base, alpha=0.12026.06 | 59.66 | 76.56 | 1,127 | |
| PowerOPDTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base, alpha=12026.06 | 59.52 | 76.86 | 1,132 | |
| PowerOPDTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base, alpha=502026.06 | 59.4 | 75.68 | 1,035 | |
| Vanilla OPD + ClipTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base2026.06 | 59.12 | 75.92 | 1,162 | |
| Vanilla OPD + TanhTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base2026.06 | 59.09 | 76.04 | 1,134 | |
| Vanilla OPDTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base2026.06 | 54.93 | 75.12 | 1,156 | |
| Vanilla OPD + Z-ScoreTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base2026.06 | 28.24 | 66.24 | 1,235 |