List your company

Mathematical Reasoning on GSM8K JSON (test)

0.854Structural AccuracyRL-Struct
0.123920.313460.5030.69254Nov 29, 2025

Evaluation results

MethodLinks
RL-StructAvg. Depth=2, learning_paradigm=Reinforcement Learning2025.110.8540.889
SFTAvg. Depth=2, learning_paradigm=Supervised Fine-Tuning2025.110.5820.658
Zero-shotAvg. Depth=2, learning_paradigm=Zero-shot2025.110.1520.352