Quantitative Reasoning on AIME
28.7AccuracyGround Truth
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Ground TruthTeacher=Qwen3-4B, Student=Qwen3-8B2026.05 | 28.7 | — | |
| Ground TruthTeacher=Qwen3-8B, Student=Qwen3-8B2026.05 | 28.4 | — | |
| NTFTeacher=Qwen3-4B, Student=Qwen3-8B2026.05 | 27.9 | 91.7 | |
| NTFTeacher=Qwen3-8B, Student=Qwen3-8B2026.05 | 27.4 | 89.2 | |
| Ground TruthTeacher=Qwen3-1.7B, Student=Qwen3-8B2026.05 | 27.4 | — | |
| I-ConfidenceTeacher=Qwen3-8B, Student=Qwen3-8B2026.05 | 27.3 | 88.4 | |
| NaiveTeacher=Qwen3-4B, Student=Qwen3-8B2026.05 | 27.2 | 84.7 | |
| V-ConfidenceTeacher=Qwen3-4B, Student=Qwen3-8B2026.05 | 27.1 | 83.7 | |
| NaiveTeacher=Qwen3-8B, Student=Qwen3-8B2026.05 | 26.8 | 83.2 | |
| I-ConfidenceTeacher=Qwen3-4B, Student=Qwen3-8B2026.05 | 26.7 | 79.6 | |
| NTFTeacher=Qwen3-1.7B, Student=Qwen3-8B2026.05 | 26.6 | 91 | |
| V-ConfidenceTeacher=Qwen3-8B, Student=Qwen3-8B2026.05 | 26.5 | 80 | |
| NaiveTeacher=Qwen3-1.7B, Student=Qwen3-8B2026.05 | 26 | 83.5 | |
| V-ConfidenceTeacher=Qwen3-1.7B, Student=Qwen3-8B2026.05 | 25.9 | 82.4 | |
| I-ConfidenceTeacher=Qwen3-1.7B, Student=Qwen3-8B2026.05 | 25.6 | 78.8 | |
| Ground TruthTeacher=Qwen3-1.7B, Student=Qwen3-4B2026.05 | 22.9 | — | |
| NTFTeacher=Qwen3-1.7B, Student=Qwen3-4B2026.05 | 22 | 92.5 | |
| I-ConfidenceTeacher=Qwen3-1.7B, Student=Qwen3-4B2026.05 | 20.7 | 81.7 | |
| NaiveTeacher=Qwen3-1.7B, Student=Qwen3-4B2026.05 | 19.6 | 72.5 | |
| No-GRPOTeacher=Qwen3-1.7B, Student=Qwen3-8B2026.05 | 18.9 | — | |
| No-GRPOTeacher=Qwen3-4B, Student=Qwen3-8B2026.05 | 18.9 | — | |
| No-GRPOTeacher=Qwen3-8B, Student=Qwen3-8B2026.05 | 18.9 | — | |
| Teacher PerformanceTeacher=Qwen3-8B, Student=Qwen3-8B2026.05 | 18.88 | — | |
| V-ConfidenceTeacher=Qwen3-1.7B, Student=Qwen3-4B2026.05 | 18.7 | 65 | |
| Teacher PerformanceTeacher=Qwen3-4B, Student=Qwen3-8B2026.05 | 10.91 | — | |
| No-GRPOTeacher=Qwen3-1.7B, Student=Qwen3-4B2026.05 | 10.9 | — | |
| Teacher PerformanceTeacher=Qwen3-1.7B, Student=Qwen3-4B2026.05 | 4.27 | — | |
| Teacher PerformanceTeacher=Qwen3-1.7B, Student=Qwen3-8B2026.05 | 4.27 | — |