List your company

Multilingual Mathematical Reasoning on PolyMath (test)

20.3Accuracy (Ar)Qwen2.5-32B-Instruct
2.3086.97911.6516.321May 21, 2026

Evaluation results

MethodLinks
Qwen2.5-32B-InstructEvaluation Type=Accuracy (Acc)2026.0520.3——176.211.917.721.819.118.619.620.518.311.81919.220.9192118.518.518.4—————————————————————
LANGBackbone=Qwen2.5-32B-Instruct2026.0519.9——20.319.313.221.422.223.422.322.921.420.614.92222.723.822.821.220.921.220.9—————————————————————
Qwen2.5-32B-InstructEvaluation Type=Logical Consistency (LC)2026.0519.8——16.316.29.417.720.41918.31920.517.610.81918.720.91918.118.117.817.7—————————————————————
LC-GRPOBackbone=Qwen2.5-32B-Instruct2026.0519.4——19.820.99.921.521.82219.421.421.519.813.421.323.52121.621.720.620.420—————————————————————
Qwen2.5-7B-Instruct + mGRPOBackbone=Qwen2.5-7B-Instruct, Technique=mGRPO2026.0516.6——15.513.19.814.716.516.416.116.81615.29.914.216.1171616.416.91515.113.7————————————————————
Qwen2.5-7B-Instruct + LANGBackbone=Qwen2.5-7B-Instruct, Technique=LANG2026.0515.7——11.415.24.115.417.917.91819.617.915.37.516.117.220.417.417.41916.615.918————————————————————
Vanilla GRPOBackbone=Qwen2.5-32B-Instruct2026.0515.4——3.510.9322.322.320.22224.320.916.58.317.519.422.821.82321.919.217.6—————————————————————
Qwen2.5-7B-Instruct + LC-GRPOBackbone=Qwen2.5-7B-Instruct, Technique=LC-GRPO2026.0514.3——911.27.713.715.917.21718.517.614.26.513.416.516.614.517.218.814.814.515————————————————————
Qwen2.5-7B-Instruct + Vanilla GRPOBackbone=Qwen2.5-7B-Instruct, Technique=Vanilla GRPO2026.0514——15.714.76.513.618.416.615.915.717.414.84.314.515.915.714.217.116.214.114.514.9————————————————————
Qwen2.5-7B-Instruct + M-ThinkerBackbone=Qwen2.5-7B-Instruct, Technique=M-Thinker2026.0513.6——8.810.75.91213.913.615.314.715.112.37.811.612.314.913.714.515.71312.613.1————————————————————
Qwen2.5-7B-Instruct + QRTBackbone=Qwen2.5-7B-Instruct, Technique=QRT2026.0512.8——11.512.24.914.515.614.916.115.117.513.57.712.31413.613.715.116.51313.311.4————————————————————
Qwen2.5-7B-InstructParameters=7B2026.0512.5——1213.3513.817.416.215.415.717.313.96.212.514.513.413.914.815.713.113.513.8————————————————————
Qwen2.5-7B-Instruct + DITBackbone=Qwen2.5-7B-Instruct, Technique=DIT2026.0512.5——10.811.83.713.216.113.616.715.816.613.16.611.611.916.813.514.914.412.612.910.9————————————————————
Qwen2.5-3B-Instruct + QRTBackbone=Qwen2.5-3B-Instruct, Technique=QRT2026.0510.8——3.26.41.69.27.79.61110.812.78.31.67.511.410.710.59.79.78.78.58.7————————————————————
Qwen2.5-3B-Instruct + LC-GRPOBackbone=Qwen2.5-3B-Instruct, Technique=LC-GRPO2026.059.9——6.610.639.711.89.911.710.212.49.638.510.510.91211.5119.69.69.3————————————————————
Qwen2.5-3B-Instruct + Vanilla GRPOBackbone=Qwen2.5-3B-Instruct, Technique=Vanilla GRPO2026.059.1——10.18.12.88.512.110.511.210.612.79.64.610.39.78.112.210.611.69.79.610.2————————————————————
Qwen2.5-7B-Instruct + LCPBackbone=Qwen2.5-7B-Instruct, Technique=LCP2026.059.1——5.18.31.38.48.38.99.310.914.28.419.49.210.1810.710.78.68.510————————————————————
Qwen2.5-3B-Instruct + mGRPOBackbone=Qwen2.5-3B-Instruct, Technique=mGRPO2026.058.9——5.98.71.58.310.41011.99.112.78.73.188.610.388.710.28.18.58.2————————————————————
Qwen2.5-3B-Instruct + LANGBackbone=Qwen2.5-3B-Instruct, Technique=LANG2026.058.9——7.410.82.79.915.610.911.211.613.310.24.59.612.211.810.512.31110.210.29.4————————————————————
Qwen2.5-3B-InstructParameters=3B2026.058.4——4.57.61.38.69.49.39.68.911.47.918.49.9109.711.811.48.78.27.1————————————————————
Vanilla GRPOBackbone=Llama3.1-8B-Instruct2026.058.3——6.27.76.17.15.98.49.6910.67.96.66.58.18.97.499.787.9—————————————————————
Qwen2.5-3B-Instruct + M-ThinkerBackbone=Qwen2.5-3B-Instruct, Technique=M-Thinker2026.058.3——4.37.71.47.6128.58.69.912.98.12.88.510.18.98.69.79.38.68.310.5————————————————————
LANGBackbone=Llama3.1-8B-Instruct2026.058.2——7.37.86.58.95.19.610.91113.48.95.66.98.59.79.310.89.88.78.8—————————————————————
Qwen2.5-3B-Instruct + LCPBackbone=Qwen2.5-3B-Instruct, Technique=LCP2026.058.2——3.78.43.46.8119.99.38.914.18.42.67.49.29.69.811.19.98.28.36.3————————————————————
Qwen2.5-3B-Instruct + DITBackbone=Qwen2.5-3B-Instruct, Technique=DIT2026.057.6——5.46.43.17.410.411.79.910.411.78.418.28.89.69.410.510.28.48.49.4————————————————————
Qwen2.5-3B-Instruct + M-SFTBackbone=Qwen2.5-3B-Instruct, Technique=M-SFT2026.057——3.57.92.15.710.68.810.27.713.87.71.48.39.510.88.910.610.68.68.18.6————————————————————
Llama3.1-8B-InstructEvaluation Type=Logical Consistency (LC)2026.055.9——5.65.95.16.94.35.876.310.36.356.27.47.46.35.86.36.36.3—————————————————————
Llama3.1-8B-InstructEvaluation Type=Accuracy (Acc)2026.055.9——6.57.16.36.97.16.67.77.310.37.25.26.87.58.17.16.36.36.87—————————————————————
Qwen2.5-7B-Instruct + M-SFTBackbone=Qwen2.5-7B-Instruct, Technique=M-SFT2026.054.5——4.26.71.74.84.24.15.64.46.14.61.46.66.57.14.97.54.75.34.94————————————————————
LC-GRPOBackbone=Llama3.1-8B-Instruct2026.053——5.87.34.77.44.79.75.18.57.86.45.56.24.642.95.364.95.8—————————————————————
Qwen2.5-3B-InstructBackbone=Qwen2.5-3B-Instruct2026.05———————————————————————8.34.57.51.38.57.99.29.68.711.47.718.49.99.59.611.411.28.78.1
Qwen2.5-3B-Instruct + DITBackbone=Qwen2.5-3B-Instruct, Variant=DIT2026.05———————————————————————7.65.36.43.17.49.211.79.910.311.78.318.28.89.59.210.210.28.28.2
Qwen2.5-3B-Instruct + LANGBackbone=Qwen2.5-3B-Instruct, Variant=LANG2026.05———————————————————————8.97.410.82.79.913.510.911.111.613.3104.59.612.211.810.512.110.910.210.1
Qwen2.5-3B-Instruct + LC-GRPOBackbone=Qwen2.5-3B-Instruct, Variant=LC-GRPO2026.05———————————————————————9.96.610.539.710.89.911.710.212.49.538.510.510.811.911.510.99.69.5
Qwen2.5-3B-Instruct + LCPBackbone=Qwen2.5-3B-Instruct, Variant=LCP2026.05———————————————————————8.23.68.43.46.810.29.99.28.814.18.32.47.49.29.59.810.99.98.48.3
Qwen2.5-3B-Instruct + M-SFTBackbone=Qwen2.5-3B-Instruct, Variant=M-SFT2026.05———————————————————————2.72.84.30.92.74.632.95.94.53.40.455.43.12.23.75.73.63.5
Qwen2.5-3B-Instruct + M-ThinkerBackbone=Qwen2.5-3B-Instruct, Variant=M-Thinker2026.05———————————————————————9.14.86.52.37.38.79.78.310.712.581.18.18.99.47.59.89.27.77.9
Qwen2.5-3B-Instruct + mGRPOBackbone=Qwen2.5-3B-Instruct, Variant=mGRPO2026.05———————————————————————8.75.88.717.58.31011.7912.78.43.17.88.610.188.410.188.2
Qwen2.5-3B-Instruct + QRTBackbone=Qwen2.5-3B-Instruct, Variant=QRT2026.05———————————————————————10.83.26.31.59.26.99.510.910.812.78.21.57.511.410.610.59.49.78.78.4
Qwen2.5-3B-Instruct + Vanilla GRPOBackbone=Qwen2.5-3B-Instruct, Variant=Vanilla GRPO2026.05———————————————————————9.10.48.11.58.510.710.211.110.612.78.32.68.59.77.812.21011.68.98.6
Qwen2.5-7B-InstructBackbone=Qwen2.5-7B-Instruct2026.05———————————————————————12.111.812.72.213.815.315.315.214.817.313.16.212.414.413.49.713.615.11212.7
Qwen2.5-7B-Instruct + DITBackbone=Qwen2.5-7B-Instruct, Variant=DIT2026.05———————————————————————12.110.811.82.813.213.713.516.615.716.612.76.611.611.916.813.514.414.312.712.7
Qwen2.5-7B-Instruct + LANGBackbone=Qwen2.5-7B-Instruct, Variant=LANG2026.05———————————————————————16.21115.33.415.315.717.917.919.417.9157.616.117.220.517.417.71916.515.6
Qwen2.5-7B-Instruct + LC-GRPOBackbone=Qwen2.5-7B-Instruct, Variant=LC-GRPO2026.05———————————————————————14.3911.22.513.713.417.21718.317.613.4613.316.516.614.517.118.714.713.9
Qwen2.5-7B-Instruct + LCPBackbone=Qwen2.5-7B-Instruct, Variant=LCP2026.05———————————————————————95.18.31.38.26.88.99.310.914.28.219.49.2107.910.510.78.48.3
Qwen2.5-7B-Instruct + M-SFTBackbone=Qwen2.5-7B-Instruct, Variant=M-SFT2026.05———————————————————————5.12.86.71.34.83.74.15.15.36.14.51.44.46.57.13.47.55.95.24.8
Qwen2.5-7B-Instruct + M-ThinkerBackbone=Qwen2.5-7B-Instruct, Variant=M-Thinker2026.05———————————————————————13.68.893.111.712.113.214.814.714.311.57.810.812.314.313.413.815.512.612
Qwen2.5-7B-Instruct + mGRPOBackbone=Qwen2.5-7B-Instruct, Variant=mGRPO2026.05———————————————————————14.910.611.52.712.713.113.815.416.415.312.66.512.813.815.713.316.416.513.613
Qwen2.5-7B-Instruct + QRTBackbone=Qwen2.5-7B-Instruct, Variant=QRT2026.05———————————————————————12.211.512.14.914.513.414.816.11517.513.27.711.51413.613.714.916.513.113.2
Qwen2.5-7B-Instruct + Vanilla GRPOBackbone=Qwen2.5-7B-Instruct, Variant=Vanilla GRPO2026.05———————————————————————13.14.51.42.113.115.816.615.915.317.411.54.212.415.815.614.116.416.213.512.3