github-top-code / README.md
ronantakizawa's picture
Upload README.md with huggingface_hub
7d7aaa7 verified
metadata
license: mit
task_categories:
  - text-generation
language:
  - code
tags:
  - code
  - github
  - source-code
  - trending-developers
  - software-engineering
size_categories:
  - 1M<n<10M

GitHub Top Developer Source Code

A curated dataset of 1.3M+ source code files from GitHub's top ranked developers (2015-2025).

This dataset is based on the top ranked developers from this dataset: https://huggingface.co/datasets/ronantakizawa/github-top-developers

Dataset Summary

  • 1.3M+ source code files from repositories across ~4,700 unique developers
  • 80+ programming languages included (Python, JavaScript, TypeScript, Rust, Go, C/C++, Java, and more)
  • Source code only — config files (JSON, YAML, TOML, etc.) and documentation (Markdown, TXT) are excluded
  • Permissive licenses only (MIT, Apache-2.0, BSD, ISC, etc.)
  • Rich metadata per file: repo stars, description, primary language, developer company affiliation

Schema

Each row represents a single source file:

Column Type Description
file_path string Path within the repo (e.g. src/main.py)
file_language string Language detected from file extension (e.g. Python, JavaScript)
content string Raw source code (UTF-8)
repo_name string Full repository name (owner/repo)
repo_stars int64 GitHub star count at time of collection
repo_description string Repository description
repo_primary_language string GitHub-detected primary language of the repository
developer_username string GitHub username
developer_name string Developer display name
developer_company string Company affiliation

Note on language columns: file_language is determined per-file from the file extension (e.g. a .py file is always Python). repo_primary_language is GitHub's auto-detected primary language for the entire repository. These may differ — for example, a C header file (.hC/C++ Header) in a repo that GitHub classifies as Python.

Splits

Split Description
train ~90% of repos — for training
test ~5% of repos — for evaluation
validation ~5% of repos — for hyperparameter tuning

Splits are assigned by repository (deterministic hash), so no repo appears in multiple splits. This prevents data leakage from files in the same project.

Usage

from datasets import load_dataset

# Load a specific split
train = load_dataset("ronantakizawa/github-top-code", split="train")
test = load_dataset("ronantakizawa/github-top-code", split="test")

# Filter by language
python_files = train.filter(lambda x: x["file_language"] == "Python")

# Filter by stars
popular = train.filter(lambda x: x["repo_stars"] > 1000)

# Get files from a specific developer
dev_files = train.filter(lambda x: x["developer_username"] == "torvalds")