Exporting from Ounias

Export types

Different Ounias outputs leave the platform in different forms.

Exports are intended to let analyses continue within the tools already used by a research or analytical workflow rather than requiring all downstream work to remain inside Ounias.

Datasets and derived tables

Export tabular artifacts generated during analysis.

Available derived tables can be exported from the relevant run or dataset context and used in external analytical workflows.

• Dataset data
• Generated labels
• Embedding coordinates

Plots

Plotly visualizations can be exported as image files.

Use Plotly's export controls on supported visualizations when a static figure is needed for a report, presentation, publication draft, or external workflow.

Trained models

Supervised models can be carried into model-management workflows through MLflow.

Ounias supports exporting trained supervised models to MLflow so model artifacts and their surrounding lifecycle can continue outside the Ounias workspace.

Configuring MLFlow is done through the app settings, and prefixes can be managed via the project settings.

More information on how to set up MLFlow is available at https://mlflow.org/docs/latest/ml/tracking/quickstart/

Structured outputs

Additional integrations can carry analytical summaries into other systems.

• Copyable metrics and analytical tables
• Notion export — planned

Snapshots are different

A Snapshot is a published analytical state rather than a conventional export.

Snapshots preserve selected analytical evidence and context in an immutable view-only form. They are intended for sharing, reference, and preservation rather than continued external computation.