Skip to content

Backbone workflow for ML Pipeline #3

@chadac

Description

@chadac

Think this one merits some discussion. In the past, I've broken up my pipelines into stages where each stage usually has an entrypoint named "run" or something similar, accepting a small set of parameters to configure and test each process independently from a sample notebook, but still broken up into pieces so deep dives from notebooks are simple as well. But I've found that this can be a bit confusing as well on the execution end, and it ended up being fairly restrictive when I wanted to rethink the pipeline structure.

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions