Edit a script
The federated modeling console provides the core functions of federated learning. It also integrates features from common machine learning libraries, such as TensorFlow, to offer out-of-the-box capabilities. When you create a script file, the system provides initial code based on the file type. You can use this code to develop models quickly and efficiently.
Procedure
After you create a script file, click the script file name. On the script editor page, click Input Configuration.
You can also edit the federated script in the script editor using Python. For more information about Python coding standards and encapsulated federated learning capabilities, see Federated Modeling API.
After you finish editing the script, click Save.
Before you execute the script file, configure its input and output. Different script types require different input and output configurations. For more information, see Script input/output configuration.
After you configure the input and output for the script file, click Execute to run the script. You can view the task execution results in the task list.
Script features
Input Configuration: Click to configure the input data for the script. You must configure the input data before you run the script. Otherwise, the script execution fails.
Output Configuration: Click to configure the name of the output federated table.
Save: Click to save your changes to the script.
Save As: Click to save the current script as a new file. By default, the new script is saved in the same folder as the original.
Execute: Click to run the script file. The system automatically creates a task execution record in the task list.
Script input/output configuration
Different script types require different input and output configurations:
Only scripts of the Create Federated Table, Pre-processing Rule, and Pre-processing Application types require output data configuration. The output data for these scripts is saved in the federated table folder.
Script type | Input configuration | Output configuration |
Create Federated Table | Source samples (multiple) | Federated table (one or more) |
Pre-processing Rule | Federated table (one or more) | Federated table (one or more) |
Pre-processing Application |
| Federated table (one or more) |
Model Development | Federated table (one or more) | Not required |
Model Evaluation |
| No configuration is required. |
When executed, scripts of the Pre-processing Application and Model Development types produce pre-processing rules and federated models. You can save these rules and models from the task execution results.