List your company

Mathematical Reasoning on GSM8K (Multilingual Breakdown)

51.7Accuracy (ar)RA-MoE
1.88414.81727.7540.683May 27, 2026

Evaluation results

MethodLinks
RA-MoEModel=Qwen1.52026.0551.732.559.259.145.860.3
RISEModel=Qwen1.52026.0550.130.657.557.743.459
SFTModel=Qwen1.52026.0548.829.556.757.242.358.5
RA-MoEModel=DSV22026.053920.765.259.951.664.3
RISEModel=DSV22026.0537.919.964.458.950.363.6
SFTModel=DSV22026.0537.619.464.258.65063.4
RSModel=Qwen1.52026.0529.54.129.441.39.945.1
0-shotModel=Qwen1.52026.0529.33.228.841.89.444.4
RSModel=DSV22026.0526.411.663.159.344.261.7
0-shotModel=DSV22026.052610.263.758.743.761.4
RA-MoEModel=OLMoE2026.0510.42.222.526.515.229.2
RISEModel=OLMoE2026.059.7221.725.714.228.1
SFTModel=OLMoE2026.059.42.221.325.713.828.3
RSModel=OLMoE2026.055.1218.621.36.823.6
0-shotModel=OLMoE2026.053.81.917.720.77.223