List your company

Mathematical Reasoning on MATH500 (Acc., Len., #Relf./Ans.)

94.8AccuracyPLE
58.545667.957877.3786.7822Apr 29, 2026

Evaluation results

MethodLinks
PLEModel=Qwen3-4B, Mode=Think2026.0494.86,050.453.56
HybridModel=Qwen3-4B, Mode=Think2026.0492.024,679.5419.6
SFT-onlyModel=Qwen2.5-7B, Mode=Think2026.0486.54,488.433.04
PLEModel=Qwen2.5-7B, Mode=Think2026.04863,861.3424.64
PLEModel=Qwen2.5-7B, Mode=No-think2026.0483.2614.460.01
HybridModel=Qwen3-4B, Mode=No-think2026.0482.221,004.130.12
PLEModel=Qwen3-4B, Mode=No-think2026.0480.8676.780.13
SFT-onlyModel=Qwen2.5-7B, Mode=No-think2026.0463.56593.090.31
InstructModel=Qwen2.5-7B, Mode=No-think2026.0459.94703.110