The Stack Metadata: Repository File Paths and Licenses for Code Datasets

Metadata for repositories used in The Stack v1.1 and v1.2 code datasets. It contains file paths and detected licenses for the source code repositories. The dataset was created by BigCode and last updated on March 16, 2023.

Use Cases

• Filtering code by license type based on the detected license information.
• Analyzing repository structure and file organization using the provided file paths.
• Recreating or verifying the composition of The Stack dataset using the metadata.
• Studying the distribution of software licenses across a large corpus of open-source code.

Strengths

• Metadata is explicitly matched to specific releases of The Stack (v1.1 and v1.2).
• Created by BigCode, a known organization in the AI-for-code research community.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Last updated 2023-03-16 13:58:24; freshness should be verified.

Provenance

Source

bigcode

Collection Method

Likely extracted from repository metadata during the creation of The Stack dataset.

Freshness

Last updated 2023-03-16 13:58:24.