View source on GitHub
|
Module for neural network layers.
Modules
base module: Contains the Template and Layer API for Oryx.
combinator module: Contains combinator layers.
convolution module: Contains building blocks for convolutional neural networks.
core module: Contains important layers for neural network construction.
normalization module: Contains building blocks for normalization layers.
pooling module: Contains building blocks for pooling layers used for neural networks.
reshape module: Contains layers that reshape arrays.
Classes
class AvgPooling: Average pooling layer, computes the average within the window.
class BatchNorm: Layer for Batch Normalization.
class Conv: Neural network layer for 2D convolution.
class Deconv: Neural network layer for 2D transposed convolution.
class Dense: Dense layer used for building neural networks.
class Dropout: Dropout layer used for building neural networks.
class Flatten: Flattens the inputs collapsing all ending dimensions.
class Layer: Base class for neural network layers.
class LayerParams: LayerParams holds params and info of Layers.
class LogSoftmax: Parent abstract class for activation functions.
class MaxPooling: Max pooling layer, computes the maximum within the window.
class Relu: Parent abstract class for activation functions.
class Reshape: Reshape the inputs to a new compatatible shape.
class Serial: Layer that executes a sequence of child layers.
class Softmax: Parent abstract class for activation functions.
class Softplus: Parent abstract class for activation functions.
class SumPooling: Sum pooling layer, computes the sum within the window.
class Tanh: Parent abstract class for activation functions.
class Template: Template class used by neural network layers.