core
====

.. py:module:: core


Classes
-------

.. autoapisummary::

   core.Unet


Functions
---------

.. autoapisummary::

   core.build_Unet


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

.. py:function:: build_Unet(input_shape, n_c_fltr=128, k_size=(3, 3), p_size=(2, 2), act_reg=None, act_reg_factor=0.0001)

   Build the Unet.

   Construct the architecture of the Unet model based on the given parameters.

   :param input_shape: The shape of the input (n_lat, n_lon, n_in_vars), where n_lat is the
                       number of latitude grid cells, n_lon is the number of longitude grid
                       cells, and n_in_vars is the number of input variables.
   :type input_shape: `tuple` of `int`
   :param n_c_fltr: The number of filters for the initial convolutional layer.
   :type n_c_fltr: `int`, optional
   :param k_size: The kernel size to use in the convolutional layers.
   :type k_size: `tuple` of `int`, optional
   :param p_size: The factors by which to downscale in the Max Pooling layers.
   :type p_size: `tuple` of `int`, optional
   :param act_reg: The name of the regularization function (of type keras.regularizers.Regularizer) to apply to the activations of the layers.
   :type act_reg: `str`, optional
   :param act_reg_factor: The regularization factor to apply to the activations of the layers.
   :type act_reg_factor: `float`, optional

   :returns: **model** -- The constructed Unet model.
   :rtype: `keras.Model`


.. py:class:: Unet

   .. py:attribute:: model
      :value: None



   .. py:method:: build(input_shape, **kwargs)


   .. py:method:: compile(optimizer, loss, **kwargs)


   .. py:method:: info()


   .. py:method:: train(*args, **kwargs)


   .. py:method:: predict(x)


   .. py:method:: summary()


   .. py:method:: load_weights(filename)


   .. py:method:: save_model(modelname)


   .. py:method:: get_config()