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

Hendrycks MATH

Benchmarks

Task NameDataset NameSOTA ResultTrend
Mathematical ReasoningHendrycks-MATH held-out 1,000-question
Accuracy (Base)71.7
9
Mathematical ReasoningHendrycks MATH
Peak Accuracy36.7
5
Math ReasoningHendrycks MATH
Pass@161.4
3
Showing 3 of 3 rows