BinaryOperator

public class BinaryOperator: ComputableOperator

Abstract class define the standard binary operator working flow. This class should not be used directly. Any class inheritance this class is doing computation on exactly two input tensors in element-wise way and return one result tensor, i.e. x + y ->>> z This BinaryOperator does not support broadcasting

attributes

operatorLabel
Operator label. Conforms to ComputableOperator
Declaration
Swift

public var operatorLabel: String = ""
Show on GitHub
metalKernelFuncLabel
This operator does not operator on GPU. Conforms to ComputableOperator
Declaration
Swift

public var metalKernelFuncLabel:String
Show on GitHub
computationDelegate
Conforms to ComputableOperator
Declaration
Swift

public var computationDelegate: OperatorCalculationDelegate?
Show on GitHub
inputTensors
Conforms to ComputableOperator
Declaration
Swift

public var inputTensors: [Tensor]?
Show on GitHub
outputTensors
Conforms to ComputableOperator
Declaration
Swift

public var outputTensors: [Tensor]?
Show on GitHub
cpuElementComputationBlock
The element compuation block in CPU mode. In most cases, subclass should just override this part in init method instead overriding the whole cpu(inputTensors:[Tensor], resultTensor: Tensor) method. The fisr tensor is input tensor A; the seconds tensor is input tensor B; the third tensor is output tensor C. This block should do some computation and assign value back to result tensor’s reader
Declaration
Swift

public var cpuElementComputationBlock: (Tensor, Tensor, Tensor) -> Void
Show on GitHub
gradComputationBlock
The grad compuation block. parameter: inputA, inputB, mode returns: An array of tensor. Should just have 2 object corresponding to two inputs
Declaration
Swift

public var gradComputationBlock: (Tensor, Tensor, OperatorComputationMode) -> [Tensor]
Show on GitHub
disableUpGradShapeCheck
If true, operator will not check the upGrads‘s shape. This is used inside framework to speed up in situation we know it will not be wrong. Cases like auto generated differentiation graph.
Declaration
Swift

public var disableUpGradShapeCheck: Bool = false
Show on GitHub
disableInputOutputCheck
If true, operator will not call inputOutputTensorsCheck() before doing calculation. This is used inside framework to speed up in situation we know it will not be wrong.
Declaration
Swift

public var disableInputOutputCheck: Bool = false
Show on GitHub
trainable
Indicate if this operator would do paramter update.

Note
All UnaryOperators are not trainable.
Declaration
Swift

public var trainable: Bool = false
Show on GitHub
mapType
The mapping type of this operator. Constant for this operator.
Declaration
Swift

public var mapType: OperatorMappingType
Show on GitHub
inPlaceble
Binary operator can do in-place calculation
Declaration
Swift

public var inPlaceble: Bool = true
Show on GitHub

Initializers


                    
                    
                    init(operatorLabel:cpuComputeBlock:gradComputationBlock:metalKernelFuncLabel:computationDelegate:)

Designated init function

Declaration

Swift

public init(operatorLabel label: String,
            cpuComputeBlock block: @escaping (Tensor, Tensor, Tensor) -> Void ,
            gradComputationBlock gradBlock: @escaping (Tensor, Tensor, OperatorComputationMode) -> [Tensor],
            metalKernelFuncLabel kernelLabel: String,
            computationDelegate: OperatorCalculationDelegate?)

Parameters

`label`	label description
`delegate`	delegate description

Show on GitHub

init(computationDelegate:)
Convenience initializer Subclass required to override this function to assign cpuComputeBlock and metalKernelFuncLabel
Declaration
Swift

required public convenience init(computationDelegate: OperatorCalculationDelegate? = nil)
Show on GitHub


                    
                    
                    init(computationDelegate:inputTensors:outputTensors:)

Convenience initializer

Declaration

Swift

public convenience init(computationDelegate: OperatorCalculationDelegate? = nil,
                        inputTensors: [Tensor], outputTensors: [Tensor])

Parameters

`computationDelegate`	computationDelegate description
`inputTensors`	inputTensors description
`outputTensors`	outputTensors description

Show on GitHub

Methods


                    
                    
                    outputShape(shapeArray:)

This operator should just receive two tensors with same dimensions (dataType could be different). Return shape is exactly the same shape as input.

Declaration

Swift

public func outputShape(shapeArray shapes: [TensorShape]) -> [TensorShape]?

Parameters


                                shapes

input shapes

Return Value

return shapes

Show on GitHub

inputOutputTensorsCheck()
The inputTensors should have exactly two tensors and same dimensions. The outputTensors should have exactly one tensora and same dimension with input tensors.
Declaration
Swift

public func inputOutputTensorsCheck() -> (check: Bool, msg: String)
Show on GitHub


                    
                    
                    computeAsync(_:)

Compute asynclly

Declaration

Swift

public func computeAsync(_ computationMode: OperatorComputationMode = SerranoEngine.configuredEngine.defaultComputationMode)

Parameters

`tensors`	input tensors
`computationMode`	computation mode

Show on GitHub


                    
                    
                    compute(_:)

Compute synclly.

Declaration

Swift

public func compute(_ computationMode: OperatorComputationMode = SerranoEngine.configuredEngine.defaultComputationMode)

Parameters

`tensors`	input tensors
`computationMode`	cmputation mode. If choose `GPU` but haven’t configued a GPU SerranoEngine, operator will use `CPU` to compute.

Return Value

result tensors

Show on GitHub


                    
                    
                    gradCompute(_:)

Calulate grads sync. All unary operator return grads tensor with same number and shape as attribute inputTensors.

Declaration

Swift

public func gradCompute(_ computationMode: OperatorComputationMode) -> [String: DataSymbolSupportedDataType]

Parameters

`computationMode`	computationMode
`upGrds`	upGrds

Return Value

return grads tensor

Show on GitHub


                    
                    
                    gradComputAsync(_:)

Cal grads async

Declaration

Swift

public func gradComputAsync(_ computationMode: OperatorComputationMode)

Parameters

`computationMode`	computationMode
`upGrds`	upGrds

Show on GitHub


                    
                    
                    updateParams(grads:LR:)

Update params if possible. No update parameters for binary operators.

Declaration

Swift

public func updateParams(grads: [Tensor], LR: Float)

Parameters

`grads`	grads tensor list
`LR`	learning rate

Show on GitHub

bindParamSymbols(_:)
Binary operator has no parameters. Do nothing
Declaration
Swift

public func bindParamSymbols(_ symbols: [GraphSymbol])
Show on GitHub
paramSymbols()
This operator has no parameters.
Declaration
Swift

public func paramSymbols() -> [GraphSymbol]
Return Value

An empty array

Show on GitHub


                    
                    
                    cpu()

Use cpu do the inplace computation. This function always do inPlace computation for inputTensorA. It’s caller function to decide the tensor’s assignment. Default, UnaryOperator defines a workflow. Subclass just needs to override cpuElementComputationBlock. If subclass needs custom flow, it could just override this function.

Note

This function should not be called from outside.

Declaration

Swift

internal func cpu()

Parameters


                                tensors

the operation tensors

Show on GitHub


                    
                    
                    gpu()

Let GPU call the Metal kernel to do the inplace computation.This function always do inPlace computation for inputTensorA. It’s caller function to decide the tensor’s assignment. Default, UnaryOperator defines a workflow. Subclass just needs to override metalKernelFuncLabel attribute. If subclass needs custom flow, it could just override this function.

Note

This function should not be called from outside.

Declaration

Swift

internal func gpu()

Parameters


                                tensors

the operation tensors

Show on GitHub