Inputs and Runs

Input Sets

The IDE treats an MMM launch as one input set:

  • config
  • dataset
  • holidays, if used

This keeps the shell aligned to how operators think about a run, even when the Python runner is still anchored through config.yml.

Loading Demo Inputs

Bundled demo data is intended for:

Inputs panel showing demo bundles and upload actions Inputs panel showing demo bundles and upload actions

  • quick evaluation of the shell
  • walkthroughs
  • smoke testing the runner and planner flow

Loading demo data copies the selected demo bundle into the open workspace so it becomes a normal input set inside that project.

Uploading Inputs

Uploading input files writes them into the workspace under a dedicated input folder. The uploaded config is then used as a normal project config candidate.

Keep dataset and holidays filenames aligned with the paths referenced in the config.

Starting A Run

When you start a run, the IDE launches the Python Abacus runner and then switches to a run-centered workflow view.

During execution:

  • the terminal carries transient process output
  • the workflow spine shows run state
  • the workbench surfaces current context and next steps

Retained Runs

A retained run is a run directory already written to disk under results/.

Retained runs matter because they let you:

  • reopen prior evidence without rerunning the model
  • inspect manifests and stage artifacts
  • continue into planning when planner-capable evidence exists

The IDE prefers retained artifact truth wherever possible.