fault_update
Prototype
def fault_update(data_path, old_model_path, new_model_path, param)
Description
Incremental training interface. It is used to update an existing model.
Working Principle
Reads data, parameters, and the existing model, performs incremental training, and returns a new model for prediction.
When using this API, ensure that the model read by this interface is the output of the fault_train or fault_update interface. Users need to ensure the security of the output to prevent attackers from triggering arbitrary code execution during model deserialization by constructing malicious data.
Parameters
Parameter |
Description |
Input/Output |
|---|---|---|
data_path |
Path to the data file to be trained |
Input |
old_model_path |
Path to the existing model file |
Input |
new_model_path |
New model file dump path |
Input |
param |
Training parameters |
Input |