TensorFloat16Bit Class

Definition

Namespace:

Windows.AI.MachineLearning

A 16-bit float tensor object.

public ref class TensorFloat16Bit sealed : ITensor, IClosable, IMemoryBuffer

/// [Windows.Foundation.Metadata.ContractVersion(Windows.AI.MachineLearning.MachineLearningContract, 65536)]
/// [Windows.Foundation.Metadata.MarshalingBehavior(Windows.Foundation.Metadata.MarshalingType.Agile)]
/// [Windows.Foundation.Metadata.Threading(Windows.Foundation.Metadata.ThreadingModel.Both)]
class TensorFloat16Bit final : ITensor, IClosable, IMemoryBuffer

[Windows.Foundation.Metadata.ContractVersion(typeof(Windows.AI.MachineLearning.MachineLearningContract), 65536)]
[Windows.Foundation.Metadata.MarshalingBehavior(Windows.Foundation.Metadata.MarshalingType.Agile)]
[Windows.Foundation.Metadata.Threading(Windows.Foundation.Metadata.ThreadingModel.Both)]
public sealed class TensorFloat16Bit : System.IDisposable, ITensor, IMemoryBuffer

Public NotInheritable Class TensorFloat16Bit
Implements IDisposable, IMemoryBuffer, ITensor

Inheritance

Object Platform::Object IInspectable TensorFloat16Bit

Attributes

ContractVersionAttribute MarshalingBehaviorAttribute ThreadingAttribute

Implements

ILearningModelFeatureValue ITensor IClosable IDisposable IMemoryBuffer

Windows requirements

Device family	Windows 10, version 1809 (introduced in 10.0.17763.0)
API contract	Windows.AI.MachineLearning.MachineLearningContract (introduced in v1.0)

Remarks

A tensor is a multi-dimensional array of values. A 16-bit float tensor is a tensor of 16-bit floating point values.

The layout of tensors is row-major, with tightly packed contiguous data representing each dimension. The total size of a tensor is the product of the size of each dimension.

Windows Server

To use this API on Windows Server, you must use Windows Server 2019 with Desktop Experience.

Thread safety

This API is thread-safe.

Version history

Windows version	SDK version	Value added
1903	18362	CreateFromBuffer
1903	18362	CreateFromShapeArrayAndDataArray

Properties

Kind	The kind of the feature.
Shape	Returns the count and size of each dimension.
TensorKind	Returns the data type of the tensor.

Methods

Close()	If there are no outstanding IMemoryBufferReference objects created from this object, Close will dispose of the object and destroy its associated resources. Otherwise, Close will detach itself from its associated resources and they will be destroyed when the last IMemoryBufferReference is closed.
Create()	Creates a 16-bit float tensor object without allocating a buffer.
Create(IIterable<Int64>)	Creates a 16-bit float tensor object and allocates a buffer of size shape.
CreateFromArray(IIterable<Int64>, Single[])	Creates a 16-bit float tensor object, allocates a buffer of size shape, and copies all of data into it.
CreateFromBuffer(Int64[], IBuffer)	Creates a 16-bit float tensor object with the given shape and uses the underlying buffer in data for subsequent evaluation.
CreateFromIterable(IIterable<Int64>, IIterable<Single>)	Creates a 16-bit float tensor object, allocates a buffer of size shape, and copies all of data into it.
CreateFromShapeArrayAndDataArray(Int64[], Single[])	Creates a 16-bit float tensor object, allocates a buffer of size shape, and copies all of data into it.
CreateReference()	Returns an IMemoryBufferReference representation of the buffer backing the 16-bit float tensor object.
Dispose()	Performs application-defined tasks associated with freeing, releasing, or resetting unmanaged resources.
GetAsVectorView()	Returns a read-only view of the data.

See also

• Windows ML
• Windows ML samples (GitHub)