Package mklab.JGNN.nn.operations
Class Reshape
java.lang.Object
mklab.JGNN.nn.NNOperation
mklab.JGNN.nn.operations.Reshape
Implements a
NNOperation
that reshapes a matrix.- Author:
- Emmanouil Krasanakis
-
Nested Class Summary
Nested classes/interfaces inherited from class mklab.JGNN.nn.NNOperation
NNOperation.ThreadData
-
Field Summary
Fields inherited from class mklab.JGNN.nn.NNOperation
debugging
-
Constructor Summary
-
Method Summary
Modifier and TypeMethodDescriptionprotected Tensor
Provides a simple description to show when drawing .dot format diagrams.boolean
Checks whether the operation's output should be cached given that it is a constant.protected Tensor
setDimensionName
(String rowName, String colName) Methods inherited from class mklab.JGNN.nn.NNOperation
addInput, autosize, clearPrediction, data, describe, getDescription, getInputs, getLastTapeError, getNonLinearity, getOutputs, getPrediction, isConstant, isInputNeededForDerivative, isOutputNeededForDerivative, run, run, runPrediction, runPredictionAndAutosize, setDescription, trainParameters, view
-
Constructor Details
-
Reshape
public Reshape(long rows, long cols)
-
-
Method Details
-
forward
- Specified by:
forward
in classNNOperation
-
getSimpleDescription
Description copied from class:NNOperation
Provides a simple description to show when drawing .dot format diagrams.- Overrides:
getSimpleDescription
in classNNOperation
- Returns:
- A string description, usually the component's class name.
-
partial
- Specified by:
partial
in classNNOperation
-
isCachable
public boolean isCachable()Description copied from class:NNOperation
Checks whether the operation's output should be cached given that it is a constant. This returnsfalse
only for randomized components that yield different outputs from different inputs, such as dropouts.- Overrides:
isCachable
in classNNOperation
- Returns:
- A
boolean
values.
-
setDimensionName
-