Mathematical Reasoning on MATH 500 (L1-L5 Metrics)
93.02Accuracy (L1)ImitKD
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ImitKDTeacher=Qwen3-32B, Student=Qwen3-4B-Instruct2026.05 | 93.02 | 82.22 | 87.62 | 77.34 | 60.45 | |
| MOTABTeacher=Qwen3-32B, Student=Qwen3-4B-Instruct2026.05 | 93.02 | 86.67 | 95.24 | 86.72 | 73.88 | |
| MOTABTeacher=gpt-oss-120b, Student=Qwen3-4B-Instruct2026.05 | 93.02 | 95.56 | 88.57 | 82.81 | 76.87 | |
| SFTTeacher=Qwen3-32B, Student=Qwen3-4B-Instruct2026.05 | 90.7 | 82.22 | 77.14 | 70.31 | 51.49 | |
| SFTTeacher=gpt-oss-120b, Student=Qwen3-4B-Instruct2026.05 | 90.7 | 90 | 80 | 74.22 | 50.75 | |
| SKDTeacher=gpt-oss-120b, Student=Qwen3-4B-Instruct, beta=0.82026.05 | 90.7 | 92.22 | 84.76 | 81.25 | 74.63 | |
| SKDTeacher=Qwen3-32B, Student=Qwen3-4B-Instruct, beta=1.02026.05 | 90.07 | 86.67 | 88.57 | 81.25 | 56.72 | |
| SKDTeacher=Qwen3-32B, Student=Qwen3-4B-Instruct, beta=0.62026.05 | 88.37 | 85.55 | 89.52 | 75.78 | 60.45 | |
| SKDTeacher=Qwen3-32B, Student=Qwen3-4B-Instruct, beta=0.82026.05 | 88.37 | 86.67 | 87.61 | 82.03 | 63.43 | |
| RFTTeacher=Qwen3-32B, Student=Qwen3-4B-Instruct2026.05 | 83.72 | 82.22 | 85.71 | 65.62 | 51.49 | |
| ImitKDTeacher=gpt-oss-120b, Student=Qwen3-4B-Instruct2026.05 | 81.4 | 67.78 | 71.43 | 69.53 | 61.19 | |
| SKDTeacher=gpt-oss-120b, Student=Qwen3-4B-Instruct, beta=1.02026.05 | 74.42 | 72.22 | 71.43 | 75 | 73.13 | |
| SKDTeacher=gpt-oss-120b, Student=Qwen3-4B-Instruct, beta=0.62026.05 | 69.77 | 74.44 | 77.14 | 71.88 | 72.39 |