---
language:
- en
tags:
- math
- mathematics
- reasoning
- arithmetic
- algebra
- word-problems
- logic
- chain-of-thought
- step-by-step
- education
- synthetic
- llm
- llm-training
- pretraining
- finetuning
- Math
- Logic
- Algebra
pretty_name: MathReasoning-2250000
size_categories:
- 1M<n<10M
license: mit
---

# MathReasoning-2250000

## A synthetic **math reasoning dataset** containing **2250000 examples**.

## Each row follows a strict reasoning-friendly structure:
- `question`
- `problem`
- `how_to_solve`
- `answer`

This format is designed for:
# - LLM reasoning
# - chain-of-thought style supervision
# - math finetuning
# - educational tasks

---

## 📐 Problem Types
# - Addition
# - Subtraction
# - Multiplication
# - Division (integer-safe)

---

## 🧪 Recommended Splits
# - Train: 80%
# - Test: 20%

---
## Generated fully in-memory. No external data sources.