Feedforward Layers
==============================================

Gluon provides a large number of build-in feedforward neural network layers in
two modiules, ``mxnet.gluon.nn`` and ``mxnet.gluon.contrib.nn``.

.. currentmodule:: mxnet.gluon


Basic Layers
------------

.. autosummary::
    :nosignatures:
    :toctree: _autogen

    nn.Dense
    nn.Activation
    nn.Dropout
    nn.Flatten
    nn.Lambda
    nn.HybridLambda

Convolutional Layers
--------------------

.. autosummary::
    :nosignatures:
    :toctree: _autogen

    nn.Conv1D
    nn.Conv2D
    nn.Conv3D
    nn.Conv1DTranspose
    nn.Conv2DTranspose
    nn.Conv3DTranspose

Pooling Layers
--------------

.. autosummary::
   :nosignatures:
   :toctree: _autogen

    nn.MaxPool1D
    nn.MaxPool2D
    nn.MaxPool3D
    nn.AvgPool1D
    nn.AvgPool2D
    nn.AvgPool3D
    nn.GlobalMaxPool1D
    nn.GlobalMaxPool2D
    nn.GlobalMaxPool3D
    nn.GlobalAvgPool1D
    nn.GlobalAvgPool2D
    nn.GlobalAvgPool3D
    nn.ReflectionPad2D

Normalization Layers
--------------------

.. autosummary::
    :nosignatures:
    :toctree: _autogen

    nn.BatchNorm
    nn.InstanceNorm
    nn.LayerNorm
    contrib.nn.SyncBatchNorm

Embedding Layers
----------------

.. autosummary::
    :nosignatures:
    :toctree: _autogen

    nn.Embedding
    contrib.nn.SparseEmbedding


Advanced Activation Layers
--------------------------

.. autosummary::
    :nosignatures:
    :toctree: _autogen

    nn.LeakyReLU
    nn.PReLU
    nn.ELU
    nn.SELU
    nn.Swish