ronantakizawa/github-top-code

---
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 (`.h` → `C/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

```python
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")
```