google-deepmind/mathematics_dataset resource

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

1965 stars · 275 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

A Python library from DeepMind that generates synthetic mathematical question-and-answer pairs at roughly school-level difficulty across topics like algebra, arithmetic, calculus, and probability. It was created to benchmark the mathematical reasoning abilities of neural models, with pre-generated data available for download.

Use cases

• generate math question-answer pairs for training models
• benchmark a model's algebraic reasoning skills
• create curriculum learning data with easy/medium/hard splits
• get synthetic arithmetic and calculus word problems
• evaluate neural networks on school-level mathematics
• generate probability and polynomial questions programmatically

When to choose

• you need large volumes of synthetic math QA data for training or evaluating models
• you want curriculum-style difficulty splits (train-easy/medium/hard)
• you need reproducible school-level math questions across many topics

When to avoid

• you need real-world or competition-level math problems
• you want a static benchmark dataset rather than a generator
• you need advanced topics beyond school-level math like topology or statistics

Facets

dataset · maturity maintenance

data-generation math machine-learning mathematics machine-learning education python cross-platform synthetic-data mathematical-reasoning question-answering benchmark school-level-math datasets

Member repositories

For agents

markdown · JSON · MCP: product_card(name="google-deepmind/mathematics_dataset")