Share your thoughts, 1 month free Claude Pro on usSee more

Mathematical Reasoning on GSM8K 8B Instruct (test)

58.6Accuracy

Prob Margin + Both

Updated 4d ago

Evaluation Results

Method	Links
Prob Margin + Both 2026.04		58.6
Top1 Prob + Both 2026.04		56.8
Top1 Prob + Planner 2026.04		55
Init. Position 2026.04		52.6
Top1 Prob + EOS Temp. 2026.04		50.9
Prob Margin 2026.04		47.2
Top1 Prob 2026.04		46.6
Temperature Sampling 2026.04		45
Ancestral Sampling 2026.04		42.1