Class CTCLossDescriptor

Inheritance

System.Object

CTCLossDescriptor

Implements

System.IDisposable

Inherited Members

System.Object.Equals(System.Object)

System.Object.Equals(System.Object, System.Object)

System.Object.GetHashCode()

System.Object.GetType()

System.Object.MemberwiseClone()

System.Object.ReferenceEquals(System.Object, System.Object)

System.Object.ToString()

Namespace: ManagedCuda.CudaDNN

Assembly: CudaDNN.dll

Syntax

public class CTCLossDescriptor : IDisposable

Constructors

CTCLossDescriptor(CudaDNNContext)

Declaration

public CTCLossDescriptor(CudaDNNContext context)

Parameters

Type	Name	Description
CudaDNNContext	context

Properties

Desc

Returns the inner handle.

Declaration

public cudnnCTCLossDescriptor Desc { get; }

Property Value

Type	Description
cudnnCTCLossDescriptor

Methods

CTCLoss(CudaDNNContext, TensorDescriptor, CudaDeviceVariable<Double>, Int32[], Int32[], Int32[], CudaDeviceVariable<Double>, TensorDescriptor, CudaDeviceVariable<Double>, cudnnCTCLossAlgo, CudaDeviceVariable<Byte>)

This function returns the ctc costs and gradients, given the probabilities and labels.

Declaration

public void CTCLoss(CudaDNNContext handle, TensorDescriptor probsDesc, CudaDeviceVariable<double> probs, int[] labels, int[] labelLengths, int[] inputLengths, CudaDeviceVariable<double> costs, TensorDescriptor gradientsDesc, CudaDeviceVariable<double> gradients, cudnnCTCLossAlgo algo, CudaDeviceVariable<byte> workspace)

Parameters

Type	Name	Description
CudaDNNContext	handle	Handle to a previously created cuDNN context.
TensorDescriptor	probsDesc	Handle to the previously initialized probabilities tensor descriptor.
CudaDeviceVariable<System.Double>	probs	Pointer to a previously initialized probabilities tensor.
System.Int32[]	labels	Pointer to a previously initialized labels list.
System.Int32[]	labelLengths	Pointer to a previously initialized lengths list, to walk the above labels list.
System.Int32[]	inputLengths	Pointer to a previously initialized list of the lengths of the timing steps in each batch.
CudaDeviceVariable<System.Double>	costs	Pointer to the computed costs of CTC.
TensorDescriptor	gradientsDesc	Handle to a previously initialized gradients tensor descriptor.
CudaDeviceVariable<System.Double>	gradients	Pointer to the computed gradients of CTC.
cudnnCTCLossAlgo	algo	Enumerant that specifies the chosen CTC loss algorithm.
CudaDeviceVariable<System.Byte>	workspace	Pointer to GPU memory of a workspace needed to able to execute the specified algorithm.

CTCLoss(CudaDNNContext, TensorDescriptor, CudaDeviceVariable<Single>, Int32[], Int32[], Int32[], CudaDeviceVariable<Single>, TensorDescriptor, CudaDeviceVariable<Single>, cudnnCTCLossAlgo, CudaDeviceVariable<Byte>)

This function returns the ctc costs and gradients, given the probabilities and labels.

Declaration

public void CTCLoss(CudaDNNContext handle, TensorDescriptor probsDesc, CudaDeviceVariable<float> probs, int[] labels, int[] labelLengths, int[] inputLengths, CudaDeviceVariable<float> costs, TensorDescriptor gradientsDesc, CudaDeviceVariable<float> gradients, cudnnCTCLossAlgo algo, CudaDeviceVariable<byte> workspace)

Parameters

Type	Name	Description
CudaDNNContext	handle	Handle to a previously created cuDNN context.
TensorDescriptor	probsDesc	Handle to the previously initialized probabilities tensor descriptor.
CudaDeviceVariable<System.Single>	probs	Pointer to a previously initialized probabilities tensor.
System.Int32[]	labels	Pointer to a previously initialized labels list.
System.Int32[]	labelLengths	Pointer to a previously initialized lengths list, to walk the above labels list.
System.Int32[]	inputLengths	Pointer to a previously initialized list of the lengths of the timing steps in each batch.
CudaDeviceVariable<System.Single>	costs	Pointer to the computed costs of CTC.
TensorDescriptor	gradientsDesc	Handle to a previously initialized gradients tensor descriptor.
CudaDeviceVariable<System.Single>	gradients	Pointer to the computed gradients of CTC.
cudnnCTCLossAlgo	algo	Enumerant that specifies the chosen CTC loss algorithm.
CudaDeviceVariable<System.Byte>	workspace	Pointer to GPU memory of a workspace needed to able to execute the specified algorithm.

CTCLoss(CudaDNNContext, TensorDescriptor, Int32[], Int32[], Int32[], TensorDescriptor, cudnnCTCLossAlgo)

return the workspace size needed for ctc

Declaration

public SizeT CTCLoss(CudaDNNContext handle, TensorDescriptor probsDesc, int[] labels, int[] labelLengths, int[] inputLengths, TensorDescriptor gradientsDesc, cudnnCTCLossAlgo algo)

Parameters

Type	Name	Description
CudaDNNContext	handle	Handle to a previously created cuDNN context.
TensorDescriptor	probsDesc	Handle to the previously initialized probabilities tensor descriptor.
System.Int32[]	labels	Pointer to a previously initialized labels list.
System.Int32[]	labelLengths	Pointer to a previously initialized lengths list, to walk the above labels list.
System.Int32[]	inputLengths	Pointer to a previously initialized list of the lengths of the timing steps in each batch.
TensorDescriptor	gradientsDesc	Handle to a previously initialized gradients tensor descriptor.
cudnnCTCLossAlgo	algo	Enumerant that specifies the chosen CTC loss algorithm

Returns

Type	Description
SizeT	Amount of GPU memory needed as workspace to be able to execute the CTC loss computation with the specified algo.

Dispose()

Dispose

Declaration

public void Dispose()

Dispose(Boolean)

For IDisposable

Declaration

protected virtual void Dispose(bool fDisposing)

Parameters

Type	Name	Description
System.Boolean	fDisposing

Finalize()

For dispose

Declaration

protected void Finalize()

SetCTCLossDescriptor(cudnnDataType)

This function initializes a previously created CTC Loss descriptor object.

Declaration

public void SetCTCLossDescriptor(cudnnDataType dataType)

Parameters

Type	Name	Description
cudnnDataType	dataType	Math precision.

Implements

System.IDisposable