List your company

Question Answering on MMLU STEM (Acc, Len, Refl/Ans)

92.22AccuracyPLE
58.128866.979475.8384.6806Apr 29, 2026

Evaluation results

MethodLinks
PLEBase Model=Qwen2.5-7B, Mode=No-think2026.0492.221390
PLEBase Model=Qwen3-4B, Mode=Think2026.0492.221,0410
HybridBase Model=Qwen3-4B, Mode=Think2026.0490.262,33816.82
PLEBase Model=Qwen2.5-7B, Mode=Think2026.0488.894290
SFT-onlyBase Model=Qwen2.5-7B, Mode=Think2026.0485.823,10024.74
HybridBase Model=Qwen3-4B, Mode=No-think2026.0484.496120.01
PLEBase Model=Qwen3-4B, Mode=No-think2026.0483.72530.01
InstructBase Model=Qwen2.5-7B, Mode=No-think2026.04622110
SFT-onlyBase Model=Qwen2.5-7B, Mode=No-think2026.0459.447740.83