Code Clippy: GitHub Source Code Dataset

16 terabytes of uncompressed source code data across 22 programming languages and 23 file extensions. The collection originates from the public GitHub dataset on Google BigQuery and targets large-scale code modeling tasks.

Use Cases

• Train large-scale code generation models using the 16 TB of raw source code
• Develop language-specific syntax highlighters for the 22 included programming languages
• Analyze coding patterns across the 23 different file extensions

Strengths

• 16 TB of uncompressed source code data
• Covers 22 programming languages
• Includes 23 unique file extensions
• Sourced from the Google BigQuery public GitHub dataset