simple-math / README.md
PingVortex's picture
Update README.md
c1a87b9 verified
|
raw
history blame
2.11 kB
metadata
license: mit
task_categories:
  - question-answering
tags:
  - math
pretty_name: Easy Math Dataset
size_categories:
  - 1M<n<10M

Beginner Math Dataset

License: MIT

A large-scale dataset containing 1,000,000 math problems and solutions for basic arithmetic operations and word problems.

Dataset Details

  • License: MIT
  • Task Category: Question Answering
  • Tags: Math, Arithmetic, Word Problems
  • Size: 100K < n < 1M (423,810 rows)
  • Split: Train only (1M rows)

Dataset Structure

Each example contains two fields:

  • input: Math problem in natural language or numerical form
  • response: Numerical answer (may include negative numbers and decimals)

Example rows:

{
    "input": "If you have one hundred and sixty apples and found four more, how many apples do you have?",
    "response": "40.0"
},
{
    "input": "976-454",
    "response": "522"
},
{
    "input": "Calculate: nineteen times twenty three",
    "response": "437"
}

Usage

Load with Hugging Face datasets:

from datasets import load_dataset

dataset = load_dataset("PingVortex/beginner-math-dataset")

Features

  • Mixed format problems (word-based and numerical)
  • Basic operations (addition, subtraction, multiplication, division)
  • Negative number calculations
  • Word problems with real-world contexts
  • Both integer and decimal solutions

Community

Join my Discord for support and updates:
Discord (Server is not invalid)

Example Use Cases

  • Math problem solving models
  • Arithmetic accuracy benchmarks
  • Word problem understanding tasks
  • Educational AI systems

Citation

If you use this dataset in your research, please credit:

@misc{easymath2025,
  author = {PingVortex},
  title = {Easy Math Dataset},
  year = {2025},
  publisher = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/PingVortex/beginner-math-dataset}}
}