List your company

Reasoning on GSM8K (Diversity and Pass@K Metrics)

0.26Distinct-1GRPO-likelihood
0.08320.12910.1750.2209Sep 5, 2025

Evaluation results

MethodLinks
GRPO-likelihoodobjective=likelihood2025.090.260.860.530.5950.980.4
GRPO-entropyobjective=entropy2025.090.10.380.120.257792.6
DQO2025.090.10.420.140.3176.391.2
GRPO2025.090.090.320.090.2176.887.9