NuminaMath-1.5: Strictly Filtered Mathematical Proof Problems

A strictly filtered subset of the NuminaMath-1.5 dataset containing only validated mathematical proof problems. The dataset was created by author 'nlile' and last updated on July 11, 2025. It applies multiple validation filters to the original data to ensure problem and solution validity.

Use Cases

• Training language models for mathematical proof generation based on validated proof problems.
• Benchmarking automated theorem provers based on the strict validation criteria.
• Analyzing the structure and complexity of mathematical proofs based on the filtered content.

Strengths

• Applies strict validation filters for 'problem_is_valid' and 'solution_is_valid'.
• Derived from an original dataset containing 110,998 rows, indicating a substantial source.
• Focuses exclusively on problems where the answer is a 'proof'.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the filtered version is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

huggingface

Collection Method

Filtered from the NuminaMath-1.5-proofs-only dataset based on validation criteria.

Time Range

null

Freshness

Last updated 2025-07-11 21:34:19; freshness should be verified.

Geography

null