NuminaMath
Fresh
An 860k-problem competition-math dataset with detailed solutions, the open community's go-to corpus for training math-specialized LLMs.
- Type
- SFT Dataset
- Publisher
- Numina
- Capabilities
- MathScientific Reasoning
- Runtime
hf_parquet- License
- Apache-2.0
- Size
- 860k problem-solution pairs (CoT variant), 73k tool-integrated (TIR)
- Published
- May 2026
Cite
Notes
Only stored in your browser.
Lift evidence
4| Eval | Tools known to lift | Source paper |
|---|---|---|
| AIME 2024: Problems from the American Invitational Mathematics Examination | NuminaMath | - |
| MATH-500 | NuminaMath | - |
| GSM8K | NuminaMath | - |
| MathVista | NuminaMath | - |
Models
Notable models trained on it
NuminaMath-7B-TIR (won AIMO Progress Prize 1, 2024)DeepSeek-Math derivativesQwen2.5-Math fine-tunescomponent of OpenThoughts and many reasoning mixtures