---
dataset_info:
  features:
  - name: source
    dtype: string
  - name: language
    dtype: string
  - name: content
    dtype: string
  - name: tokens
    dtype: int64
  - name: metadata
    struct:
    - name: type
      dtype: string
    - name: id
      dtype: string
  splits:
  - name: train
    num_bytes: 127466907
    num_examples: 59244
  download_size: 24864346
  dataset_size: 127466907
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
description: null
license: mit
language:
- en
tags:
- code
size_categories:
- 1M<n<10M
---
🚀 MUD-Code3-Mega – 10M Tokens of High‑Quality Code
This dataset contains 59244 synthetic code snippets designed to mimic real‑world production‑grade code across multiple domains (ML, web, async, data processing, deep learning, etc.). All samples are carefully crafted to be realistic, well‑structured, and high‑quality.

Total tokens: 10,000,112
Languages: Python (with some snippets including other languages like SQL, Dockerfile)
Quality: High – generated from expert‑level templates with variations.
📖 How to Use
from datasets import load_dataset
dataset = load_dataset("CompiwerAI/MUD-Code3-Mega")
print(dataset["train"][0])

📊 Stats
Metric	Value
Total Documents	59244
Total Tokens	10,000,112
File Size (raw)	~129 MB
🔍 Why This Dataset?
🧠 Large scale – 10M tokens for robust training.
🧪 High quality – templates from production code patterns.
🌍 Diverse – covers many domains and paradigms.
📦 Ready to use – standard format, no preprocessing needed.
📜 License & Citation
MIT License.
If you use this dataset, please cite:

@misc{mud-code3-mega-2026,
  author = {CompiwerAI},
  title = {MUD‑Code3-Mega: A 10M‑token High‑Quality Code Dataset},
  year = {2026},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/CompiwerAI/MUD-Code3-Mega}
}