List your company

Mathematical Reasoning on GSM8K (accuracy (G8))

91.7Accuracy (GSM8K)GRPO
36.16450.5826579.418Dec 30, 2025

Evaluation results

MethodLinks
GRPOModel Size=7B2025.1291.7
iCLPModel Size=7B2025.1290.1
FT-SModel Size=7B2025.1288.9
FT-EModel Size=7B2025.1287.3
GRPOModel Size=3B2025.1286.6
BaseModel Size=7B2025.1285.4
iCLPModel Size=3B2025.1285
FT-SModel Size=3B2025.1283.2
FT-EModel Size=3B2025.1281.4
BaseModel Size=3B2025.1279.1
GRPOModel Size=0.5B2025.1254.5
iCLPModel Size=0.5B2025.1251.5
FT-SModel Size=0.5B2025.1248
FT-EModel Size=0.5B2025.1243.8
BaseModel Size=0.5B2025.1241.6
FT-LEModel Size=0.5B2025.1238.3