phitodeep.model
===============

.. py:module:: phitodeep.model


Classes
-------

.. autoapisummary::

   phitodeep.model.Sequential
   phitodeep.model.SequentialBuilder


Module Contents
---------------

.. py:class:: Sequential(*layers, alpha=0.01, optimizer: phitodeep.optimization.optimizers.Optimizer = o.Adam(), batch_size=None, epochs=1000, loss_class=ls.MeanSquaredError())

   .. py:attribute:: layers
      :value: []



   .. py:attribute:: alpha
      :value: 0.01



   .. py:attribute:: optimizer_type


   .. py:attribute:: batch_size
      :value: None



   .. py:attribute:: epochs
      :value: 1000



   .. py:attribute:: loss_class


   .. py:method:: add(layer) -> None

      Add a layer to the network.



   .. py:method:: setoptimizer(optimizer)


   .. py:method:: setbatchsize(num)


   .. py:method:: setloss(loss_class)


   .. py:method:: train(X, y, X_test, y_test)

      Train the model using the specified optimizer and loss function.

      :param X: Training data.
      :type X: np.ndarray
      :param y: Training labels.
      :type y: np.ndarray
      :param X_test: Test data.
      :type X_test: np.ndarray
      :param y_test: Test labels.
      :type y_test: np.ndarray

      :returns: A list of tuples containing the training and test losses for each epoch.
      :rtype: list



   .. py:method:: predict(X)

      Forward pass through all layers.

      :param X: input array

      :returns: output after passing through all layers



   .. py:method:: backward(gradient)

      Backward pass through all layers.

      :param gradient: dL/dY from loss function (shape: batch_size x output_size)

      Propagates gradient backwards through all layers in reverse order.
      Each layer computes its parameter gradients, updates parameters,
      and returns the gradient for the previous layer.



   .. py:method:: __call__(X)

      Allow model(X) syntax.



   .. py:method:: summary()

      Print model architecture.



   .. py:method:: copy()

      Return a copy of the model.



.. py:class:: SequentialBuilder

   Fluent API for building Sequential models.


   .. py:attribute:: layers
      :value: []



   .. py:attribute:: alpha_value
      :value: 1



   .. py:attribute:: optimizer_type


   .. py:attribute:: batch_size
      :value: None



   .. py:attribute:: epochs_value
      :value: 1000



   .. py:attribute:: loss_class


   .. py:method:: flatten()

      Add a Flatten layer.



   .. py:method:: dense(input_size, output_size, initializer=He())

      Add a Dense layer.



   .. py:method:: relu()

      Add a ReLU activation.



   .. py:method:: sigmoid()

      Add a Sigmoid activation.



   .. py:method:: tanh()

      Add a Tanh activation.



   .. py:method:: softmax()

      Add a Softmax activation.



   .. py:method:: elu(alpha_activation=1.0)

      Add an ELU activation.



   .. py:method:: optimizer(optimizer)

      Set the optimizer.



   .. py:method:: batch(num)

      Set the batch size.



   .. py:method:: alpha(num)

      Set the learning rate.



   .. py:method:: epochs(num)

      Set the number of epochs.



   .. py:method:: loss(loss_class)

      Set the loss function.



   .. py:method:: build()

      Build and return the Sequential model.