gen-release-notes

To automate PyTorch release notes generation, use a frontmatter-driven workflow that transforms scattered PR data into a publish-ready worksheet for a specified version and functional area. It validates inputs and organizes entries into versioned todo and done files.

Organizing scattered PyTorch PR data into a release notes worksheet requires a structured workflow that validates inputs, categorizes entries by functional area, and tracks miscategorized items. This produces versioned todo and done markdown files ready for finalization.

Managing miscategorized PR entries when compiling PyTorch release notes involves an optional tracking mechanism within the generation workflow. It identifies and handles entries assigned to incorrect functional areas before finalizing the versioned done file.

The PyTorch release notes workflow fully supports versioned todo and done file structures. It maps generated notes directly into the repository's release notes layout for a specified version, ensuring seamless integration with existing markdown categorization workflows.

You can generate release notes for any specified PyTorch version and supported functional area combination, such as aotdispatcher, dynamo, or inductor. The workflow requires both parameters to validate inputs and produce the correct categorized markdown output.