UGMathBench: A Diverse and Dynamic Benchmark for Undergraduate-Level Mathematical Reasoning with Large Language Models

UGMathBench is a diverse and dynamic benchmark specifically designed for evaluating undergraduate-level mathematical reasoning with LLMs. UGMathBench comprises 5,062 problems across 16 subjects and 111 topics, featuring 10 distinct answer types. Each problem includes three randomized versions.

An Example to load the data

from datasets import load_dataset
dataset=load_dataset("UGMathBench/ugmathbench", "Trigonometry", split="test")

print(dataset[0])

More details on loading and using the data are on our GitHub page.

If you do find our code helpful or use our benchmark dataset, please cite our paper.

@article{xu2025ugmathbench,
  title={UGMathBench: A Diverse and Dynamic Benchmark for Undergraduate-Level Mathematical Reasoning with Large Language Models},
  author={Xu, Xin and Zhang, Jiaxin and Chen, Tianhao and Chao, Zitong and Hu, Jishan and Yang, Can},
  journal={arXiv preprint arXiv:2501.13766},
  year={2025}
}