msp.layers package

Module contents

The mps.layers module includes utility to create different types of tensorflow custom layers to define MSP model architecture.

class msp.layers.ContextEmbedding(*args, **kwargs)[source]

Bases: keras.engine.base_layer.Layer

build(input_shape)[source]

Create the state of the layer (weights)

call(inputs)[source]
class msp.layers.GGCNLayer(*args, **kwargs)[source]

Bases: keras.engine.base_layer.Layer

build(input_shape)[source]

Create the state of the layer (weights)

call(inputs, training=None)[source]
class msp.layers.MHALayer(*args, **kwargs)[source]

Bases: keras.engine.base_layer.Layer

build(input_shape)[source]

Create the state of the layer (weights)

call(inputs, training=None)[source]