metadata.json

{
    "dataset_name": "Nightglow Code Dataset",
    "version": "1.0.0",
    "languages": ["python", "javascript", "java"],
    "total_samples": 100000,
    "sources": ["github", "stackoverflow", "leetcode"],
    "created_date": "2026-07-03",
    "preprocessing": {
        "cleaned": true,
        "deduplicated": true,
        "filtered": true
    }
}