BenchmarkMATH数据集 / MATH benchmark
MATH Benchmark
A mathematical reasoning benchmark introduced by Hendrycks et al. in 2021, comprising 12,500 competition-level mathematics problems spanning seven mathematical domains. It is widely regarded as a standard evaluation dataset for assessing the mathematical capabilities of AI models.
Core Facts
Timeline (last 90 days)
Aug 8
The MATH benchmark contains competition-level mathematics problems, covering 7 areas including algebra, geometry, and number theory, for a total of 12,500 problems.
Verified65%
Source Articles
Noam Brown: AI Math Capabilities Growing 10x Per Year, Exceeding ExpectationsSep 23
DeepSeek Math V2深度解析：开源数学AI的能力天花板Jul 24