SciCode Domain Code: 1.1 Billion Lines of Code from Scientific GitHub Repositories

DATA1 is a large-scale domain-specific code dataset containing over 1.1 billion lines of code. It was collected from GitHub repositories by SciCodePile and covers 178 interdisciplinary topics in fields like biology, chemistry, and materials science. The dataset was last updated on March 13, 2026.

Use Cases

• Train or fine-tune large language models for code generation based on domain-specific scientific code.
• Analyze coding patterns and practices within interdisciplinary scientific fields mentioned in the description.
• Develop tools for automated documentation or bug detection in scientific software repositories.

Strengths

• Large scale with over 1.1 billion lines of code.
• Broad coverage of 178 different domain-specific topics.
• Sourced from real-world GitHub repositories, suggesting practical relevance.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Description metadata is limited; actual data quality requires manual inspection after download.
• The dataset's composition and potential biases inherent to its GitHub source are not detailed.

Provenance

Source

GitHub repositories.

Collection Method

Collected and organized from GitHub.

Freshness

Last updated 2026-03-13 05:19:22; freshness should be verified.