MATH Benchmark

What is MATH?

MATH is a new dataset of 12,500 challenging competition mathematics problems. Each problem in MATH has a full step-by-step solution.

Released

2021-11-08

Evaluates

数学, Math, 大语言模型, LLM, 数理能力, 开源收录, Open-Source, 支持, Supported

Openness

open

Importer

OpenCompass Hub

Reported scores

0

Source provenance

• Original evidence https://hub.opencompass.org.cn/dataset-detail/MATH

MATH paper, code and dataset

• Paper https://arxiv.org/pdf/2103.03874.pdf
• Code https://github.com/hendrycks/math
• Dataset https://huggingface.co/datasets/hendrycks/competition_math

Which sources cite MATH?

• MATH

Related 数学, Math, 大语言模型, LLM, 数理能力, 开源收录, Open-Source, 支持, Supported benchmarks

• AbsPyramid
• Ada-LEval
• AFQMC
• AGIEval
• AIDABench
• AIR-Bench
• AIRBench-2024
• ALE-Bench