The model wrapper uses a combination of the service_type and service_name fields in the input JSON to select the appropriate model for a given request.
When a request is received, the wrapper first looks at the service_type to determine whether the request is for property prediction or data generation.
For property prediction, the wrapper uses the service_type (e.g., get_molecule_property, get_protein_property, get_crystal_property) and the service_name to look up the appropriate predictor in the PropertyPredictorRegistry.
For data generation, the wrapper uses the service_type (generate_data) and the service_name to look up the appropriate generator in the GeneratorRegistry.
The model wrapper can be configured to use different models for the same service_type and service_name based on the other parameters in the request. This is achieved by registering multiple models with the same service_type and service_name but with different parameter schemas.
For example, you could have two different models for predicting the solubility property of a molecule, one that uses a simple linear regression model and another that uses a more complex deep learning model. The wrapper would select the appropriate model based on the other parameters in the request, such as the model_type parameter.