Tensor.SoftMax<T> Method

Applies the SoftMax function along the specified axis of a tensor.

Definition

Namespace: Numerics.NET.Tensors
Assembly: Numerics.NET.Tensors (in Numerics.NET.Tensors.dll) Version: 9.0.3

public static Tensor<T> SoftMax<T>(
	Tensor<T> tensor,
	Index axis,
	Tensor<T>? result = null,
	Tensor<bool>? mask = null,
	bool skipMissingValues = false,
	TensorElementOrder order = TensorElementOrder.Automatic
)

Visual Basic

Public Shared Function SoftMax(Of T) ( 
	tensor As Tensor(Of T),
	axis As Index,
	Optional result As Tensor(Of T) = Nothing,
	Optional mask As Tensor(Of Boolean) = Nothing,
	Optional skipMissingValues As Boolean = false,
	Optional order As TensorElementOrder = TensorElementOrder.Automatic
) As Tensor(Of T)

C++/CLI

public:
generic<typename T>
static Tensor<T>^ SoftMax(
	Tensor<T>^ tensor, 
	Index axis, 
	Tensor<T>^ result = nullptr, 
	Tensor<bool>^ mask = nullptr, 
	bool skipMissingValues = false, 
	TensorElementOrder order = TensorElementOrder::Automatic
)

static member SoftMax : 
        tensor : Tensor<'T> * 
        axis : Index * 
        ?result : Tensor<'T> * 
        ?mask : Tensor<bool> * 
        ?skipMissingValues : bool * 
        ?order : TensorElementOrder 
(* Defaults:
        let _result = defaultArg result null
        let _mask = defaultArg mask null
        let _skipMissingValues = defaultArg skipMissingValues false
        let _order = defaultArg order TensorElementOrder.Automatic
*)
-> Tensor<'T>

Parameters

tensor Tensor<T>: A tensor.
axis Index: The axis along which the transformation is computed.
result Tensor<T> (Optional): Optional. The tensor that is to hold the result. May be null.
mask Tensor<Boolean> (Optional): Optional. A boolean tensor that specifies for which elements the operation should be performed. If null (the default), the operation is applied everywhere.
skipMissingValues Boolean (Optional): Optional. Specifies whether missing values should be excluded from the transformation. The default is false.
order TensorElementOrder (Optional): Optional. If result is null, the desired ordering in memory of the result.

Type Parameters

T: The type of the elements of the tensor.

Return Value

Tensor<T>
A tensor of the same shape as tensor whose elements are equal to the sorted of the corresponding elements of tensor.

Exceptions

ArgumentNullException

tensor is null.