NuminaMath
860K competition and olympiad-level math problems with curated, structured solutions. Created for the AI Mathematical Olympiad challenge, NuminaMath represents the frontier of mathematical reasoning datasets, covering AMC, AIME, and international olympiad problems.
from datasets import load_dataset
ds = load_dataset("AI-MO/NuminaMath-CoT")Preview a sample row
{
"problem": "Let $a, b, c$ be positive reals with $a + b + c = 1$. Find the minimum value of $\\frac{1}{a} + \\frac{4}{b} + \\frac{9}{c}$.",
"solution": "By Cauchy-Schwarz, $(a+b+c)(\\frac{1}{a}+\\frac{4}{b}+\\frac{9}{c}) \\ge (1+2+3)^2 = 36$. Equality holds at $a=1/6, b=2/6, c=3/6$. Min = 36.",
"source": "AMC-2024"
}Fine-tune with this dataset
Estimated VRAM to fine-tune with QLoRA (4-bit base model + LoRA adapters), using conservative defaults:
New to fine-tuning? Follow the step-by-step walkthrough: Fine-Tune Your First LLM in 1 Hour
Frequently asked questions
Can I use NuminaMath commercially?
Yes — NuminaMath is released under Apache 2.0, a permissive license that allows commercial use, including training models you ship in a product. Check the dataset card for attribution requirements before release.
How much data does NuminaMath contain, and do I need all of it?
NuminaMath contains 860K Problems. You rarely need all of it: for style and format fine-tuning, a few hundred to a few thousand examples are enough — load a slice (e.g. split="train[:1000]") and scale up only if quality plateaus.
What is NuminaMath best used for?
Chain-of-thought math fine-tuning up to olympiad level. It belongs to the Reasoning section of our dataset hub, where you'll find alternatives and complementary sets.