MyCaffe.layers.tft Namespace Reference

The MyCaffe.layers.tft namespace contains all TFT related layers. More...

Classes
class	CategoricalTransformationLayer
	The CategoricalTransformationLayer implements the transforming/embeddings for the set of categorical input variables from a single input channel. Each input variable is projected using a dedicated embedding layer to a vector of width state_size. The result of applying this module is a list, with length num_inputs, that contians the embedding of each input variable for all observations and time steps. More...
class	ChannelEmbeddingLayer
	The ChannelEmbeddingLayer implements the transforming/embeddings for both the numeric and categorical data of an input channel. This layer manages both a NumericTransformationLayer and CategoricalTransformationLayer. More...
class	DataSchema
	The DataSchema class is used by the DataTemporalLayer to load the data schema describing the NPY files containing the dataset. More...
class	DataTemporalLayer
	The DataTemporalLayer implements the data layer used to load the temporal data into the model. More...
class	Field
	The Field class manages a single field. More...
class	FieldCollection
	The FieldCollection manages a collection of fields. More...
class	GateAddNormLayer
	The GateAddNormLayer implements the Dropout, Gated Linear Unit layer, LayerNorm while adding in the residual. More...
class	GluLayer
	The GluLayer implements the Gated Linear Unit layer. More...
class	GrnLayer
	The GrnLayer implements the Gated Linear Unit layer. More...
class	LayerFactory
	The LayerFactor is responsible for creating all layers implemented in the MyCaffe.layers.tft namespace. More...
class	Lookup
	The Lookup class is used to manage a single lookup table. More...
class	LookupCollection
	The LookupCollection class contains a collection of Lookup objects. More...
class	LookupItem
	The lookup item manages a single lookup item used to map a name to an ID. More...
class	MultiHeadAttentionInterpLayer
	The MultiHeadAttentionInterpLayer implements the Multi-head Attention Interpretive Layer More...
class	NumericTransformationLayer
	The NumericTransformationLayer implements the transforming/embeddings for the set of numeric input variables from a single input channel. Each input variable is projected using a dedicated inner product layer to a vector of width state_size. The result of applying this module is a list, with length num_inputs, that contians the embedding of each input variable for all observations and time steps. More...
class	QuantileAccuracyLayer
	The QuantileAccuracyLayer implements the Quantile Accuracy Layer used in TFT models. More...
class	RawData
	The RawData class is the base class for all raw data types. More...
class	RawFileData
	The RawFileData object is used to load raw NPY file data. More...
class	RawSqlData
	The RawSqlData class loads data from a database. More...
class	ReshapeTemporalLayer
	The ReshapeTemporalLayer implements the Variable Selection Network More...
class	VarSetNetLayer
	The VarSetNetLayer implements the Variable Selection Network More...

Detailed Description

The MyCaffe.layers.tft namespace contains all TFT related layers.