List your company

Multilingual Knowledge Evaluation on Global-MMLU High

68.35AccuracyGRPO
34.08242.978551.87560.7715Jun 4, 2026

Evaluation results

MethodLinks
GRPOBackbone=Qwen-2.5-7B2026.0668.35
BaselineBackbone=Qwen-2.5-7B2026.0668.15
CPT + GRPOBackbone=Qwen-2.5-7B2026.0665.2
SFTBackbone=Qwen-2.5-7B2026.0664.45
CPT + SFTBackbone=Qwen-2.5-7B2026.0664.2
CPTBackbone=Qwen-2.5-7B2026.0663.95
GRPOBackbone=OLMo-2-1124-7B2026.0639.22
BaselineBackbone=OLMo-2-1124-7B2026.0638.72
CPTBackbone=OLMo-2-1124-7B2026.0637.41
CPT + GRPOBackbone=OLMo-2-1124-7B2026.0636.34
CPT + SFTBackbone=OLMo-2-1124-7B2026.0636.15
SFTBackbone=OLMo-2-1124-7B2026.0635.4