Import Model
External models can be imported into the Data Science Lab and experimented inside the Notebooks.
Please Note:
The External models can be registered to the Data Pipeline module and they can be inferred using the Data Science Lab script runner.
Only the Native prediction functionality will work for the External models.
Importing a Model
Navigate to the Model tab for a Repo Sync Project.
Click the Import Model option.
The user gets redirected to upload the model file. Select and upload the file.
A notification message appears.
The imported model gets added to the model list.
Exporting the Model to Data Pipeline
The user needs to start a new .ipynb file with wrapper function that includes Data, Imported Model, Predict function, and output Dataset with predictions.
Create a new .ipynb file or navigate to an existing .ipynb file.
Use the code cell to write the needed wrapper function.
Access the Imported Model inside this .ipynb file.
Load it and use the model number inside your script.
Run the script.
Run the code cell with the inference script to get the preview of the data.
Register the model using the Models tab inside the .ipynb file.
The Register Model dialog box appears to confirm about the model registration.
Click the Yes option.
A notification message appears, and the model gets registered.
Export the script using the Export functionality provided for the .ipynb file.
The Export to Pipeline window appears.
Select the Export to pipeline option.
Select a specific script from the Notebook. or Choose the Select All option to select the full script.
Select the Next option.
Click the Validate icon to validate the script.
A notification message appears to assure the validity of the script.
Click the Export to Pipeline option.
A notification message appears to assure that the selected Notebook has been exported.
Please Note: The imported model gets registered to the Data Pipeline module as a script.
Accessing the Exported Model within the Pipeline User interface
Navigate to the Data Pipeline Workflow editor.
Drag the DS Lab Runner component and configure the Basic Information.
Open the Meta Information tab of the DS Lab Runner component.
Configure the following information for the Meta Information tab.
Select Script Runner as the Execution Type.
Select function input type.
Select the project name.
Select the Script Name from the drop-down option. The same name given to the imported model appears as the script name.
Provide details for the External Library (if applicable).
Select the Start Function from the drop-down menu.
The exported model along with the script can be accessed inside the Script section.
The user can connect the DS Lab Script Runner component to an Input Event.
Run the Pipeline.
The model predictions can be generated in the Preview tab of the connected Input Event.
Please Note:
The Imported Models can be accessed through the Script Runner component inside the Data Pipeline module.
The other models can be accessed through the Model Runner component inside the Data Pipeline.
Sample files for Sklearn
Sample files for Keras
Sample files for PyTorch
Please Note:
The supported extensions for External models - .pkl, .h5, .pth & .pt
Refer the Data Science Lab Quick Start Flow page to get an overview of the Data Science Lab module in nutshell.
Last updated
