Importing Models from Kaggle
The Model Import feature allows you to add custom machine learning models to your project.
While SolutionEngine includes built-in models, many use cases require specialized or domain-specific models. The Kaggle import feature provides a simple way to bring external models into the Model Library without manual packaging or file management.
Why Import Models?
Built-in models may not cover every task or dataset.
Importing from Kaggle allows you to:
- Use publicly available community-trained models
- Integrate domain-specific solutions
- Extend the platform beyond the default model set
Once imported, a Kaggle model behaves like any other model in the Model Library and can be used inside workflows.
How Kaggle Import Works
Importing a model from Kaggle requires only the model’s public URL.
To import:
- Navigate to the Model Library inside your project.
- Select Import Model.
- Paste the Kaggle model URL.
- Confirm the import.
After submission, SolutionEngine handles the remaining steps automatically.
The platform:
- Fetches the model files.
- Downloads the required weights.
- Processes the model artifacts.
- Generates the necessary execution wrapper and metadata.
Once processing is complete, the model appears in your Model Library and becomes available for use in workflows.
After Import
Imported models can be selected in workflow nodes just like built-in models.
Each model includes configurable parameters depending on its structure and capabilities.
If the import process fails due to invalid URLs or unsupported formats, an error message will be displayed with details.
Notes
- The Kaggle model must be publicly accessible.
- The platform currently supports importing from Kaggle only.
- Additional import sources may be supported in future releases.
For details on using a model inside a workflow, refer to the Workflows section.