List your company

Mathematical Reasoning on GSM8K 500-example (test)

84.8Pass@1 FinalAH-GRPO
62.85668.55374.2579.947Jun 3, 2026

Evaluation results

MethodLinks
AH-GRPOModel=3B, alpha=0.5, Steps=1502026.0684.886.20.0630.0051
GRPOModel=3B, alpha=0.0, Steps=1502026.0684.685.20.08850.0054
SA-AH-GRPOModel=3B, alpha=0.5, Steps=1802026.0684.685.80.02460.0079
BaseModel=3B, Evaluation Protocol=zero-shot2026.0683.1———
SA-AH-GRPOModel=1.5B, alpha=0.5, Steps=1802026.0668.668.60.17180.0099
GRPOModel=1.5B, alpha=0.0, Steps=1502026.066666.80.08850.0054
AH-GRPOModel=1.5B, alpha=0.5, Steps=1502026.066567.60.0630.0051
BaseModel=1.5B, Evaluation Protocol=zero-shot2026.0663.7———