Skip to main content

Neural Network Model

Build neural networks using Morph's built-in NN module.

Model Structure​

A neural network model is a class with a Forward method:

module NN;

Classifier class {
    layer1 is NN.Linear(784, 128);
    layer2 is NN.Linear(128, 10);

    Forward method(input as Tensor<float>) as Tensor<float> {
        x is NN.ReLU(layer1.Forward(input));
        return NN.Softmax(layer2.Forward(x));
    }
}

Creating a Model​

model is Classifier();
output is model.Forward(inputData);

Module Import​

module NN;

This provides access to layers, optimizers, activation functions, and the training API.

Self-Test​

Verify the NN runtime is working:

module NN;

Init method() {
    result is NN.SelfTest();
    if (result == 1) {
        Print("NN self-test passed.");
    }
}

Next Steps​