-
Notifications
You must be signed in to change notification settings - Fork 0
aurora: Model Storing
Clouke edited this page May 5, 2023
·
3 revisions
The ModelStoring supports saving all models with the Model
implementation
such as NeuralNetworkModel
, LinearRegressionModel
, etc.
In this example, we'll be storing a Neural Network
model
The model we're using in this example:
NeuralNetworkTrainer nn = new NeuralNetworkBuilder()
// required parameters, such as learning rate, epochs, layers, etc.
.build();
If the model is unnamed or already has been named, you can do the following:
Model model = nn.toModel();
Convert with a name:
Model model = nn.toModel("my_model");
Store and apply the specified directory
model.saveDirectoryPath("my_directory") // applies the save directory path to the model
.save();
Or specify the directory directly:
model.save("my_directory");
If the model already has a directory path, you can do the following:
model.save();