List your company

Mathematical Reasoning on GSM8K (Avg@8, Pass@8, Avg Response Length)

83.5Avg@8Full-vocab KL OPD
41.296852.253463.2174.1666Jun 15, 2026

Evaluation results

MethodLinks
Full-vocab KL OPDTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base2026.0683.5953,954
PowerOPDTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base, alpha=5002026.0683.1394.31316
PowerOPDTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base, alpha=502026.0683.1194.24313
PowerOPDTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base, alpha=102026.0683.0194.69315
PowerOPDTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base, alpha=1002026.0682.9293.33307
PowerOPDTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base, alpha=52026.0682.6895.07315
PowerOPDTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base, alpha=0.12026.0682.4694.77316
Vanilla OPD + TanhTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base2026.0682.0894.62331
PowerOPDTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base, alpha=12026.0682.0495.38317
Vanilla OPD + ClipTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base2026.0680.3694.01326
Vanilla OPDTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base2026.0673.9395.68351
Vanilla OPD + Z-ScoreTeacher=Qwen3-4B, Student=Qwen3-1.7B-Base2026.0642.9289.61372