Hugging Face's logo

Datasets:
brando
/
putnam_bench_informal

Modalities:
Text
Formats:
parquet
Size:
< 1K
ArXiv:
Libraries:
Datasets
pandas
Dataset card Viewer Files Community
1

You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

By requesting access to this dataset, you agree to cite the following work in any publications or projects that utilize this data:

  • PutnamBench dataset: @misc{tsoukalas2024putnambenchevaluatingneuraltheoremprovers,
    title={PutnamBench: Evaluating Neural Theorem-Provers on the Putnam Mathematical Competition},
    author={George Tsoukalas and Jasper Lee and John Jennings and Jimmy Xin and Michelle Ding and Michael Jennings and Amitayush Thakur and Swarat Chaudhuri},
    year={2024},
    eprint={2407.11214},
    archivePrefix={arXiv},
    primaryClass={cs.AI},
    url={https://arxiv.org/abs/2407.11214}

Log in or Sign Up to review the conditions and access this dataset content.

Note: this is not my work, just uploading for convinience, please cite them.

Dataset Card for PutnamBench

Dataset Description

PutnamBench is a benchmark designed to evaluate theorem-proving algorithms on mathematics problems from the William Lowell Putnam Mathematical Competition (1962–2023). This dataset provides multilingual support for three formal languages: Lean 4, Isabelle, and Coq. It includes 1696 manually crafted formalizations, aggregated across all languages. This HF card is only for the informal stuff hosted at: https://github.com/trishullab/PutnamBench/blob/main/informal/putnam.json

Supported Formal Languages (not in this HF card)

PutnamBench includes formalizations in:

  • Lean 4: 644 problems
  • Isabelle: 640 problems
  • Coq: 412 problems

Statistics

The dataset includes: - Algebra: 253 problems - Analysis: 226 problems - Number Theory: 108 problems - Geometry: 69 problems - Linear Algebra: 51 problems - Abstract Algebra: 28 problems - Combinatorics: 29 problems - Probability: 10 problems - Set Theory: 8 problems

Formal Leaderboard

They are hosting a leaderboard on PutnamBench to share evaluation results. To submit results, please include a link to a preprint or publication. For private submissions, reach out to george.tsoukalas@utexas.edu.

Citation

If you use this dataset in your research, please cite:

@misc{tsoukalas2024putnambenchevaluatingneuraltheoremprovers,
      title={PutnamBench: Evaluating Neural Theorem-Provers on the Putnam Mathematical Competition}, 
      author={George Tsoukalas and Jasper Lee and John Jennings and Jimmy Xin and Michelle Ding and Michael Jennings and Amitayush Thakur and Swarat Chaudhuri},
      year={2024},
      eprint={2407.11214},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2407.11214}, 
}

arxiv: https://arxiv.org/abs/2407.11214

Natural Language (Informal) Putnam Benchmark

If you are interested in an natural language automatically evaluatable benchmark, see our Putnam-AXIOM data set and it's functional variations: https://huggingface.co/datasets/Putnam-AXIOM/putnam-axiom-dataset

Downloads last month
35