List your company

Mathematical Reasoning on GSM8K p

99.2Accuracyentropy-guided decoding
96.49697.19897.998.602Mar 10, 2026

Evaluation results

MethodLinks
entropy-guided decodingModel=Llama3.1-8B2026.0399.23,4854,3230.05
entropy-guided decodingModel=GPT52026.0396.61255080.52