Migrating a DS Lab Model via Git

The Model Migration feature simplifies collaboration by allowing you to export a Data Science Lab (DSL) model to a Git repository and import it into a different tenant or account. This process avoids duplication, ensures consistency, and helps you maintain a clear version history, ultimately accelerating teamwork and streamlining workflows across different environments.

Note:

• The DS Lab module appears as Dslabs in the Git migration interface.

• Model Migration does not support transfers within the same space. Import must target a different space (or a different tenant).

Pre-requisites for Migration

Before migrating any DSL script or model, complete the DS Lab Migration configuration in the Admin module, including:

• Version Control Settings

• Execution Settings

• DS Lab Settings

Export a DSL Model (to Git)

Use this to migrate a model from your current space to Git so it can be imported elsewhere.

Navigation path: Data Science Lab (DS Lab) > Models > Model Migration

Steps

1. Open the Models list from the DS Lab module.

2. Select a model from the list.

3. Click on the selected model to get the actions drawer for the model.

4. Click the Model Migration icon for that model.

   • The Export to GIT dialog opens.

5. Enter a Commit Message.

6. Click Yes to confirm.

7. A notification confirms the model has been exported (migrated) to Git.

Tip: Keep commit messages descriptive (e.g., “v1.2 – tuned thresholds, updated metadata”) to simplify version tracking during import.

Import a DSL Model (from Git)

Navigation path: Admin Module > Git Migration > Import File > Import Document > Migration – Document Import

1. Navigate to the Admin Module.

2. Click the Git Migration icon from the side navigation bar.

3. Click Import File.

4. On the Import Document page, click Import.

5. The Document Import page opens.

6. Select New VCS for the Version Control configuration.

7. From the Select Module drop-down menu, choose Dslabs.

8. From the Migration – Document Import drop-down menu, select Models.

9. Use the Search bar to locate the specific model to import.

10. Click the model from the displayed list to select it.

11. The available versions of the selected model appear on the right side of the page.

12. Select the desired version by checking the corresponding checkbox.

13. Click Migrate.

14. A notification message confirms that the model has been successfully migrated.

Accessing the Migrated Model

1. Open the Data Science Lab module.

2. Navigate to the Models list page within the project.

3. The migrated model will appear in the list, ready for use in notebooks or pipelines.

Important:

• The associated DSL project containing the migrated model is also transferred to the target user’s account.

• Ensure that migration pre-requisites are completed to avoid errors during the import process.

Behavior & Limitations

• Spaces: Import must target a different space (or different tenant).

• Project Transfer: The associated DSL project is transferred along with the model.

• Same-Space Transfers: Not supported—use a different space or tenant.

• Auditability: Commit messages and Git history serve as the audit trail of model evolution.

Troubleshooting & Tips

• Migration option unavailable: Verify Admin → DS Lab Migration is configured.

• Import list empty: Confirm the model was exported successfully, and you’re in the correct target space/tenant. Use the Search bar with the exact model name.

• Permission errors: Ensure you have the required Admin/DS Lab privileges in both source and target environments.

• Version ambiguity: Prefer the latest stable version tag or use descriptive commit messages to avoid importing unintended versions.

• Linked assets: If the model expects specific external libraries or data paths, validate environment parity after import.