List your company

Multiple Choice Question Answering on MMLU STEM

84.91ScoreQwen2.5(14B)T
50.860459.700268.5477.3798Jul 4, 2026

Evaluation results

MethodLinks
Qwen2.5(14B)TEvaluation Protocol=Generation-based2026.0784.91——
Qwen2.5(14B)TEvaluation Protocol=Likelihood-based2026.0775.55——
RG-OPDEvaluation Protocol=Generation-based2026.0759.462.752
Qwen2.5(1.5B)SEvaluation Protocol=Generation-based2026.0756.7102.4
Reverse KLEvaluation Protocol=Generation-based2026.0756.3-0.413
RG-OPDEvaluation Protocol=Likelihood-based2026.0753.850.282
Qwen2.5(1.5B)SEvaluation Protocol=Likelihood-based2026.0753.5702.4
TSD-KDEvaluation Protocol=Likelihood-based2026.0753.44-0.132.6
TSD-KDEvaluation Protocol=Generation-based2026.0752.55-4.162.6
Reverse KLEvaluation Protocol=Likelihood-based2026.0752.17-1.43