---
license: cc-by-sa-4.0
task_categories:
- text-generation
- question-answering
language:
- en
pretty_name: AutoMathText
size_categories:
- 10B<n<100B
configs:
- config_name: web-0.50-to-1.00
  data_files: 
  - split: train
    path: 
    - data/web/0.95-1.00.jsonl
    - data/web/0.90-0.95.jsonl
    - data/web/0.85-0.90.jsonl
    - data/web/0.80-0.85.jsonl
    - data/web/0.75-0.80.jsonl
    - data/web/0.70-0.75.jsonl
    - data/web/0.65-0.70.jsonl
    - data/web/0.60-0.65.jsonl
    - data/web/0.55-0.60.jsonl
    - data/web/0.50-0.55.jsonl
  default: true
- config_name: web-0.60-to-1.00
  data_files: 
  - split: train
    path: 
    - data/web/0.95-1.00.jsonl
    - data/web/0.90-0.95.jsonl
    - data/web/0.85-0.90.jsonl
    - data/web/0.80-0.85.jsonl
    - data/web/0.75-0.80.jsonl
    - data/web/0.70-0.75.jsonl
    - data/web/0.65-0.70.jsonl
    - data/web/0.60-0.65.jsonl
- config_name: web-0.70-to-1.00
  data_files: 
  - split: train
    path: 
    - data/web/0.95-1.00.jsonl
    - data/web/0.90-0.95.jsonl
    - data/web/0.85-0.90.jsonl
    - data/web/0.80-0.85.jsonl
    - data/web/0.75-0.80.jsonl
    - data/web/0.70-0.75.jsonl
- config_name: web-0.80-to-1.00
  data_files: 
  - split: train
    path: 
    - data/web/0.95-1.00.jsonl
    - data/web/0.90-0.95.jsonl
    - data/web/0.85-0.90.jsonl
    - data/web/0.80-0.85.jsonl
- config_name: web-full
  data_files: data/web/*.jsonl
- config_name: arxiv-0.50-to-1.00
  data_files: 
  - split: train
    path:
    - data/arxiv/0.90-1.00/*.jsonl 
    - data/arxiv/0.80-0.90/*.jsonl
    - data/arxiv/0.70-0.80/*.jsonl
    - data/arxiv/0.60-0.70/*.jsonl
    - data/arxiv/0.50-0.60/*.jsonl
- config_name: arxiv-0.60-to-1.00
  data_files: 
  - split: train
    path:
    - data/arxiv/0.90-1.00/*.jsonl 
    - data/arxiv/0.80-0.90/*.jsonl
    - data/arxiv/0.70-0.80/*.jsonl
    - data/arxiv/0.60-0.70/*.jsonl
- config_name: arxiv-0.70-to-1.00
  data_files: 
  - split: train
    path:
    - data/arxiv/0.90-1.00/*.jsonl 
    - data/arxiv/0.80-0.90/*.jsonl
    - data/arxiv/0.70-0.80/*.jsonl
- config_name: arxiv-0.80-to-1.00
  data_files: 
  - split: train
    path:
    - data/arxiv/0.90-1.00/*.jsonl 
    - data/arxiv/0.80-0.90/*.jsonl
- config_name: arxiv-full
  data_files: 
  - split: train
    path: 
    - data/arxiv/0.90-1.00/*.jsonl 
    - data/arxiv/0.80-0.90/*.jsonl
    - data/arxiv/0.70-0.80/*.jsonl
    - data/arxiv/0.60-0.70/*.jsonl
    - data/arxiv/0.50-0.60/*.jsonl
    - data/arxiv/0.00-0.50/*.jsonl
- config_name: code-0.50-to-1.00
  data_files: 
  - split: train
    path: 
    - data/code/agda/0.95-1.00.jsonl
    - data/code/agda/0.90-0.95.jsonl
    - data/code/agda/0.85-0.90.jsonl
    - data/code/agda/0.80-0.85.jsonl
    - data/code/agda/0.75-0.80.jsonl
    - data/code/agda/0.70-0.75.jsonl
    - data/code/agda/0.65-0.70.jsonl
    - data/code/agda/0.60-0.65.jsonl
    - data/code/agda/0.55-0.60.jsonl
    - data/code/agda/0.50-0.55.jsonl
    - data/code/c/0.95-1.00.jsonl
    - data/code/c/0.90-0.95.jsonl
    - data/code/c/0.85-0.90.jsonl
    - data/code/c/0.80-0.85.jsonl
    - data/code/c/0.75-0.80.jsonl
    - data/code/c/0.70-0.75.jsonl
    - data/code/c/0.65-0.70.jsonl
    - data/code/c/0.60-0.65.jsonl
    - data/code/c/0.55-0.60.jsonl
    - data/code/c/0.50-0.55.jsonl
    - data/code/cpp/0.95-1.00.jsonl
    - data/code/cpp/0.90-0.95.jsonl
    - data/code/cpp/0.85-0.90.jsonl
    - data/code/cpp/0.80-0.85.jsonl
    - data/code/cpp/0.75-0.80.jsonl
    - data/code/cpp/0.70-0.75.jsonl
    - data/code/cpp/0.65-0.70.jsonl
    - data/code/cpp/0.60-0.65.jsonl
    - data/code/cpp/0.55-0.60.jsonl
    - data/code/cpp/0.50-0.55.jsonl
    - data/code/fortran/0.95-1.00.jsonl
    - data/code/fortran/0.90-0.95.jsonl
    - data/code/fortran/0.85-0.90.jsonl
    - data/code/fortran/0.80-0.85.jsonl
    - data/code/fortran/0.75-0.80.jsonl
    - data/code/fortran/0.70-0.75.jsonl
    - data/code/fortran/0.65-0.70.jsonl
    - data/code/fortran/0.60-0.65.jsonl
    - data/code/fortran/0.55-0.60.jsonl
    - data/code/fortran/0.50-0.55.jsonl
    - data/code/gap/0.95-1.00.jsonl
    - data/code/gap/0.90-0.95.jsonl
    - data/code/gap/0.85-0.90.jsonl
    - data/code/gap/0.80-0.85.jsonl
    - data/code/gap/0.75-0.80.jsonl
    - data/code/gap/0.70-0.75.jsonl
    - data/code/gap/0.65-0.70.jsonl
    - data/code/gap/0.60-0.65.jsonl
    - data/code/gap/0.55-0.60.jsonl
    - data/code/gap/0.50-0.55.jsonl
    - data/code/github-coq-train/0.95-1.00.jsonl
    - data/code/github-coq-train/0.90-0.95.jsonl
    - data/code/github-coq-train/0.85-0.90.jsonl
    - data/code/github-coq-train/0.80-0.85.jsonl
    - data/code/github-coq-train/0.75-0.80.jsonl
    - data/code/github-coq-train/0.70-0.75.jsonl
    - data/code/github-coq-train/0.65-0.70.jsonl
    - data/code/github-coq-train/0.60-0.65.jsonl
    - data/code/github-coq-train/0.55-0.60.jsonl
    - data/code/github-coq-train/0.50-0.55.jsonl
    - data/code/github-isabelle-train/0.95-1.00.jsonl
    - data/code/github-isabelle-train/0.90-0.95.jsonl
    - data/code/github-isabelle-train/0.85-0.90.jsonl
    - data/code/github-isabelle-train/0.80-0.85.jsonl
    - data/code/github-isabelle-train/0.75-0.80.jsonl
    - data/code/github-isabelle-train/0.70-0.75.jsonl
    - data/code/github-isabelle-train/0.65-0.70.jsonl
    - data/code/github-isabelle-train/0.60-0.65.jsonl
    - data/code/github-isabelle-train/0.55-0.60.jsonl
    - data/code/github-isabelle-train/0.50-0.55.jsonl
    - data/code/github-lean-train/0.95-1.00.jsonl
    - data/code/github-lean-train/0.90-0.95.jsonl
    - data/code/github-lean-train/0.85-0.90.jsonl
    - data/code/github-lean-train/0.80-0.85.jsonl
    - data/code/github-lean-train/0.75-0.80.jsonl
    - data/code/github-lean-train/0.70-0.75.jsonl
    - data/code/github-lean-train/0.65-0.70.jsonl
    - data/code/github-lean-train/0.60-0.65.jsonl
    - data/code/github-lean-train/0.55-0.60.jsonl
    - data/code/github-lean-train/0.50-0.55.jsonl
    - data/code/github-MATLAB-train/0.95-1.00.jsonl
    - data/code/github-MATLAB-train/0.90-0.95.jsonl
    - data/code/github-MATLAB-train/0.85-0.90.jsonl
    - data/code/github-MATLAB-train/0.80-0.85.jsonl
    - data/code/github-MATLAB-train/0.75-0.80.jsonl
    - data/code/github-MATLAB-train/0.70-0.75.jsonl
    - data/code/github-MATLAB-train/0.65-0.70.jsonl
    - data/code/github-MATLAB-train/0.60-0.65.jsonl
    - data/code/github-MATLAB-train/0.55-0.60.jsonl
    - data/code/github-MATLAB-train/0.50-0.55.jsonl
    - data/code/haskell/0.95-1.00.jsonl
    - data/code/haskell/0.90-0.95.jsonl
    - data/code/haskell/0.85-0.90.jsonl
    - data/code/haskell/0.80-0.85.jsonl
    - data/code/haskell/0.75-0.80.jsonl
    - data/code/haskell/0.70-0.75.jsonl
    - data/code/haskell/0.65-0.70.jsonl
    - data/code/haskell/0.60-0.65.jsonl
    - data/code/haskell/0.55-0.60.jsonl
    - data/code/haskell/0.50-0.55.jsonl
    - data/code/idris/0.95-1.00.jsonl
    - data/code/idris/0.90-0.95.jsonl
    - data/code/idris/0.85-0.90.jsonl
    - data/code/idris/0.80-0.85.jsonl
    - data/code/idris/0.75-0.80.jsonl
    - data/code/idris/0.70-0.75.jsonl
    - data/code/idris/0.65-0.70.jsonl
    - data/code/idris/0.60-0.65.jsonl
    - data/code/idris/0.55-0.60.jsonl
    - data/code/idris/0.50-0.55.jsonl
    - data/code/isa_proofsteps/0.95-1.00.jsonl
    - data/code/isa_proofsteps/0.90-0.95.jsonl
    - data/code/isa_proofsteps/0.85-0.90.jsonl
    - data/code/isa_proofsteps/0.80-0.85.jsonl
    - data/code/isa_proofsteps/0.75-0.80.jsonl
    - data/code/isa_proofsteps/0.70-0.75.jsonl
    - data/code/isa_proofsteps/0.65-0.70.jsonl
    - data/code/isa_proofsteps/0.60-0.65.jsonl
    - data/code/isa_proofsteps/0.55-0.60.jsonl
    - data/code/isa_proofsteps/0.50-0.55.jsonl
    - data/code/julia/0.95-1.00.jsonl
    - data/code/julia/0.90-0.95.jsonl
    - data/code/julia/0.85-0.90.jsonl
    - data/code/julia/0.80-0.85.jsonl
    - data/code/julia/0.75-0.80.jsonl
    - data/code/julia/0.70-0.75.jsonl
    - data/code/julia/0.65-0.70.jsonl
    - data/code/julia/0.60-0.65.jsonl
    - data/code/julia/0.55-0.60.jsonl
    - data/code/julia/0.50-0.55.jsonl
    - data/code/jupyter-notebook/0.95-1.00.jsonl
    - data/code/jupyter-notebook/0.90-0.95.jsonl
    - data/code/jupyter-notebook/0.85-0.90.jsonl
    - data/code/jupyter-notebook/0.80-0.85.jsonl
    - data/code/jupyter-notebook/0.75-0.80.jsonl
    - data/code/jupyter-notebook/0.70-0.75.jsonl
    - data/code/jupyter-notebook/0.65-0.70.jsonl
    - data/code/jupyter-notebook/0.60-0.65.jsonl
    - data/code/jupyter-notebook/0.55-0.60.jsonl
    - data/code/jupyter-notebook/0.50-0.55.jsonl
    - data/code/lean_proofsteps/0.95-1.00.jsonl
    - data/code/lean_proofsteps/0.90-0.95.jsonl
    - data/code/lean_proofsteps/0.85-0.90.jsonl
    - data/code/lean_proofsteps/0.80-0.85.jsonl
    - data/code/lean_proofsteps/0.75-0.80.jsonl
    - data/code/lean_proofsteps/0.70-0.75.jsonl
    - data/code/lean_proofsteps/0.65-0.70.jsonl
    - data/code/lean_proofsteps/0.60-0.65.jsonl
    - data/code/lean_proofsteps/0.55-0.60.jsonl
    - data/code/lean_proofsteps/0.50-0.55.jsonl
    - data/code/maple/0.95-1.00.jsonl
    - data/code/maple/0.90-0.95.jsonl
    - data/code/maple/0.85-0.90.jsonl
    - data/code/maple/0.80-0.85.jsonl
    - data/code/maple/0.75-0.80.jsonl
    - data/code/maple/0.70-0.75.jsonl
    - data/code/maple/0.65-0.70.jsonl
    - data/code/maple/0.60-0.65.jsonl
    - data/code/maple/0.55-0.60.jsonl
    - data/code/maple/0.50-0.55.jsonl
    - data/code/python/0.95-1.00.jsonl
    - data/code/python/0.90-0.95.jsonl
    - data/code/python/0.85-0.90.jsonl
    - data/code/python/0.80-0.85.jsonl
    - data/code/python/0.75-0.80.jsonl
    - data/code/python/0.70-0.75.jsonl
    - data/code/python/0.65-0.70.jsonl
    - data/code/python/0.60-0.65.jsonl
    - data/code/python/0.55-0.60.jsonl
    - data/code/python/0.50-0.55.jsonl
    - data/code/r/0.95-1.00.jsonl
    - data/code/r/0.90-0.95.jsonl
    - data/code/r/0.85-0.90.jsonl
    - data/code/r/0.80-0.85.jsonl
    - data/code/r/0.75-0.80.jsonl
    - data/code/r/0.70-0.75.jsonl
    - data/code/r/0.65-0.70.jsonl
    - data/code/r/0.60-0.65.jsonl
    - data/code/r/0.55-0.60.jsonl
    - data/code/r/0.50-0.55.jsonl
    - data/code/tex/0.95-1.00.jsonl
    - data/code/tex/0.90-0.95.jsonl
    - data/code/tex/0.85-0.90.jsonl
    - data/code/tex/0.80-0.85.jsonl
    - data/code/tex/0.75-0.80.jsonl
    - data/code/tex/0.70-0.75.jsonl
    - data/code/tex/0.65-0.70.jsonl
    - data/code/tex/0.60-0.65.jsonl
    - data/code/tex/0.55-0.60.jsonl
    - data/code/tex/0.50-0.55.jsonl
- config_name: code-python-0.50-to-1.00
  data_files: 
  - split: train
    path: 
    - data/code/python/0.95-1.00.jsonl
    - data/code/python/0.90-0.95.jsonl
    - data/code/python/0.85-0.90.jsonl
    - data/code/python/0.80-0.85.jsonl
    - data/code/python/0.75-0.80.jsonl
    - data/code/python/0.70-0.75.jsonl
    - data/code/python/0.65-0.70.jsonl
    - data/code/python/0.60-0.65.jsonl
    - data/code/python/0.55-0.60.jsonl
    - data/code/python/0.50-0.55.jsonl
- config_name: code-python-0.60-to-1.00
  data_files: 
  - split: train
    path: 
    - data/code/python/0.95-1.00.jsonl
    - data/code/python/0.90-0.95.jsonl
    - data/code/python/0.85-0.90.jsonl
    - data/code/python/0.80-0.85.jsonl
    - data/code/python/0.75-0.80.jsonl
    - data/code/python/0.70-0.75.jsonl
    - data/code/python/0.65-0.70.jsonl
    - data/code/python/0.60-0.65.jsonl
- config_name: code-python-0.70-to-1.00
  data_files: 
  - split: train
    path: 
    - data/code/python/0.95-1.00.jsonl
    - data/code/python/0.90-0.95.jsonl
    - data/code/python/0.85-0.90.jsonl
    - data/code/python/0.80-0.85.jsonl
    - data/code/python/0.75-0.80.jsonl
    - data/code/python/0.70-0.75.jsonl
- config_name: code-python-0.80-to-1.00
  data_files: 
  - split: train
    path: 
    - data/code/python/0.95-1.00.jsonl
    - data/code/python/0.90-0.95.jsonl
    - data/code/python/0.85-0.90.jsonl
    - data/code/python/0.80-0.85.jsonl
- config_name: code-jupyter-notebook-0.50-to-1.00
  data_files: 
  - split: train
    path: 
    - data/code/jupyter-notebook/0.95-1.00.jsonl
    - data/code/jupyter-notebook/0.90-0.95.jsonl
    - data/code/jupyter-notebook/0.85-0.90.jsonl
    - data/code/jupyter-notebook/0.80-0.85.jsonl
    - data/code/jupyter-notebook/0.75-0.80.jsonl
    - data/code/jupyter-notebook/0.70-0.75.jsonl
    - data/code/jupyter-notebook/0.65-0.70.jsonl
    - data/code/jupyter-notebook/0.60-0.65.jsonl
    - data/code/jupyter-notebook/0.55-0.60.jsonl
    - data/code/jupyter-notebook/0.50-0.55.jsonl
- config_name: code-jupyter-notebook-0.60-to-1.00
  data_files: 
  - split: train
    path: 
    - data/code/jupyter-notebook/0.95-1.00.jsonl
    - data/code/jupyter-notebook/0.90-0.95.jsonl
    - data/code/jupyter-notebook/0.85-0.90.jsonl
    - data/code/jupyter-notebook/0.80-0.85.jsonl
    - data/code/jupyter-notebook/0.75-0.80.jsonl
    - data/code/jupyter-notebook/0.70-0.75.jsonl
    - data/code/jupyter-notebook/0.65-0.70.jsonl
    - data/code/jupyter-notebook/0.60-0.65.jsonl
- config_name: code-jupyter-notebook-0.70-to-1.00
  data_files: 
  - split: train
    path: 
    - data/code/jupyter-notebook/0.95-1.00.jsonl
    - data/code/jupyter-notebook/0.90-0.95.jsonl
    - data/code/jupyter-notebook/0.85-0.90.jsonl
    - data/code/jupyter-notebook/0.80-0.85.jsonl
    - data/code/jupyter-notebook/0.75-0.80.jsonl
    - data/code/jupyter-notebook/0.70-0.75.jsonl
- config_name: code-jupyter-notebook-0.80-to-1.00
  data_files: 
  - split: train
    path: 
    - data/code/jupyter-notebook/0.95-1.00.jsonl
    - data/code/jupyter-notebook/0.90-0.95.jsonl
    - data/code/jupyter-notebook/0.85-0.90.jsonl
    - data/code/jupyter-notebook/0.80-0.85.jsonl
- config_name: code-full
  data_files: 
  - split: train
    path: 
    - data/code/*/*.jsonl
tags:
- mathematical-reasoning
- reasoning
- finetuning
- pretraining
- llm
---

# AutoMathText

**AutoMathText** is an extensive and carefully curated dataset encompassing around **200 GB** of mathematical texts. It's a compilation sourced from a diverse range of platforms including various websites, arXiv, and GitHub (OpenWebMath, RedPajama, Algebraic Stack). This rich repository has been **autonomously selected (labeled) by the state-of-the-art open-source language model**, Qwen-72B. Each piece of content in the dataset is assigned **a score `lm_q1q2_score` within the range of [0, 1]**, reflecting its relevance, quality and educational value in the context of mathematical intelligence.

GitHub homepage: https://github.com/yifanzhang-pro/AutoMathText

ArXiv paper: https://arxiv.org/abs/2402.07625

## Objective

The primary aim of the **AutoMathText** dataset is to provide a comprehensive and reliable resource for a wide array of users - from academic researchers and educators to AI practitioners and mathematics enthusiasts. This dataset is particularly geared towards:

- Facilitating advanced research in **the intersection of mathematics and artificial intelligence**.
- Serving as an educational tool for **learning and teaching complex mathematical concepts**.
- Providing **a foundation for developing and training AI models** specialized in processing and understanding **mathematical content**.

## Configs

```YAML
configs:
  - config_name: web-0.50-to-1.00
    data_files:
      - split: train
        path:
          - data/web/0.95-1.00.jsonl
          - data/web/0.90-0.95.jsonl
          - ...
          - data/web/0.50-0.55.jsonl
    default: true
  - config_name: web-0.60-to-1.00
  - config_name: web-0.70-to-1.00
  - config_name: web-0.80-to-1.00
  - config_name: web-full
    data_files: data/web/*.jsonl
  - config_name: arxiv-0.50-to-1.00
    data_files:
      - split: train
        path:
          - data/arxiv/0.90-1.00/*.jsonl
          - ...
          - data/arxiv/0.50-0.60/*.jsonl
  - config_name: arxiv-0.60-to-1.00
  - config_name: arxiv-0.70-to-1.00
  - config_name: arxiv-0.80-to-1.00
  - config_name: arxiv-full
    data_files: data/arxiv/*/*.jsonl
  - config_name: code-0.50-to-1.00
    data_files:
      - split: train
        path:
          - data/code/*/0.95-1.00.jsonl
          - ...
          - data/code/*/0.50-0.55.jsonl
  - config_name: code-python-0.50-to-1.00
      - split: train
        path:
          - data/code/python/0.95-1.00.jsonl
          - ...
          - data/code/python/0.50-0.55.jsonl
  - config_name: code-python-0.60-to-1.00
  - config_name: code-python-0.70-to-1.00
  - config_name: code-python-0.80-to-1.00
  - config_name: code-jupyter-notebook-0.50-to-1.00
      - split: train
        path:
          - data/code/jupyter-notebook/0.95-1.00.jsonl
          - ...
          - data/code/jupyter-notebook/0.50-0.55.jsonl
  - config_name: code-jupyter-notebook-0.60-to-1.00
  - config_name: code-jupyter-notebook-0.70-to-1.00
  - config_name: code-jupyter-notebook-0.80-to-1.00
  - config_name: code-full
    data_files: data/code/*/*.jsonl
```

How to load data:

```python
from datasets import load_dataset

ds = load_dataset("math-ai/AutoMathText", "web-0.50-to-1.00") # or any valid config_name
```

## Features

- **Volume**: Approximately 200 GB of text data (in natural language and programming language).
- **Content**: A diverse collection of mathematical texts, including but not limited to research papers, educational articles, and code documentation.
- **Labeling**: Every text is **scored** by Qwen-72B, a sophisticated language model, ensuring a high standard of relevance and accuracy.
- **Scope**: Covers a wide spectrum of mathematical topics, making it suitable for various applications in advanced research and education.

## References

- OpenWebMath [[link]](https://huggingface.co/datasets/open-web-math/open-web-math)
- RedPajama [[link]](https://huggingface.co/datasets/togethercomputer/RedPajama-Data-1T)
- Algebraick Stack [[link]](https://huggingface.co/datasets/EleutherAI/proof-pile-2) (a subset of Proof-Pile-2)

## Citation 
We appreciate your use of **AutoMathText** in your work. If you find this repository helpful, please consider citing it and star this repo. Feel free to contact zhangyif21@tsinghua.edu.cn or open an issue if you have any questions (GitHub homepage: https://github.com/yifanzhang-pro/AutoMathText).

```bibtex
@article{zhang2024automathtext,
      title={AutoMathText: Autonomous Data Selection with Language Models for Mathematical Texts},
      author={Zhang, Yifan and Luo, Yifan and Yuan, Yang and Yao, Andrew Chi-Chih},
      journal={arXiv preprint arXiv:2402.07625},
      year={2024},
}
```