Skip to content

CNTK Library Native Eval Interface

Jump to bottom
Zhou Wang edited this page Apr 2, 2017 · 6 revisions

The C++ CNTK Library for Evaluation is based on CNTK Library API. The following methods are used to evaluate a model.

The following methods load a model into Function.

static FunctionPtr CNTK::Function::LoadModel(const std::wstring& modelFile, const DeviceDescriptor& computeDevice = DeviceDescriptor::UseDefaultDevice())

This method loads a model file and returns the pointer to the Function that represents the loaded model. A Function in CNTK denotes a computation graph composed of primitive Functions or composite functions with zero or more input arguments and one or more outputs. The modelFile is the model file path. The computeDevice specifies the device to run evaluation.

static FunctionPtr CNTK::Function::LoadModel(char *modelBuffer, size_t modelBufferLength, const DeviceDescriptor& computeDevice = DeviceDescriptor::UseDefaultDevice())

This method loads a model from a memory buffer and returns the pointer to the Function that represents the loaded model. The modelBuffer points to the buffer containing the serialized model content, and the modelBufferLength is the buffer length. The computeDevice specifies the device to run evaluation.

The following method evaluates the Function using specified input.

void CNTK::Function::Evaluate(const std::unordered_map<Variable, ValuePtr>& arguments, std::unordered_map<Variable, ValuePtr>& outputs, const DeviceDescriptor& computeDevice = DeviceDescriptor::UseDefaultDevice())

This method starts evaluation of the this Function representing the model with specificed parameters. The arguments contains values of all input variables required for evaluation, and the outputs store the values of output variables. The storage of output values can either be pre-allocated by the caller, or by the system if the ValuePtr mapping is passed as null by the caller. The computeDevice specifies the device to run evaluation.

The following methods are helper funcitons to manipuate the Function to be evaluated.

FunctionPtr CNTK::Function::Clone(ParameterCloningMethod parameterCloneMethod = ParameterCloningMethod::Clone, const std::unordered_map<Variable, Variable>& replacements = {}) const;

This method clones this Function. For evaluation, this method is called to create a cloned Function which can then be used by another thread to evaluate the same model. For this purpose, the parameterCloneMethod should be set to its default value ParameterCloningMethod::Share. The parameter replacements specifies any variable replacements that are applied in the cloned Function instance, and is usually not needed for evaluation.

FunctionPtr CNTK::Function::FindByName(const std::wstring& name, bool nestedSearchInsideBlockFunction = false)

Find a function with the given name in the Function graph underlying 'this' Function. If more than one function with the same name, an exception is thrown. The paramter name specifies the name that the returned function should match. If nestedSearchInsideBlockFunction is true, all functions inside block functions are also searched for the given name.

std::vector<FunctionPtr> CNTK::Function::FindAllWithName(const std::wstring& name, bool nestedSearchInsideBlockFunction = false)

Find a list of functions with the given name in the Function graph underlying 'this' Function. The paramter name specifies the name that the returned function should match. If nestedSearchInsideBlockFunction is true, all functions inside block functions are also searched for the given name.

FunctionPtr CNTK::Combine(const std::vector<Variable>& operands, const std::wstring& name = L"")

Create a new Function instance which just combines the outputs of the specified list of _operands_ of Functions such that the 'Outputs' of the new 'Function' are union of the 'Outputs' of each of the specified operands Functions. As an example, when creating a classification model, typically the CrossEntropy loss Function and the ClassificationError Function comprise the two roots of the computation graph which can be "Combine"d to create a single Function with 2 outputs; viz. CrossEntropy loss and ClassificationError output.

FunctionPtr CNTK::AsComposite(const FunctionPtr& rootFunction, const std::wstring& name = L"")

Creates a composite Function that has the specified _rootFunction_ as its root. The composite denotes a higher-level Function encapsulating the entire graph of Functions underlying the specified rootFunction.

FunctionPtr CNTK::Alias(const Variable& operand, const std::wstring& name = L"")

Creates a new Function instance which is just an alias of the specified _operand_.

The following methods create Value object from input data in dense format.

template <typename ElementType> static ValuePtr CNTK::Value::CreateBatch(const NDShape& sampleShape, const std::vector<ElementType>& batchData, const DeviceDescriptor& device, bool readOnly = false)

Creates a new Value object containing a batch of samples. The number of samples in the batch is the number of elements in batchData divided by the size of sampleShape (A runtime error occurs if the remainder is not zero). The created Value object contains a copy of the specified data in batchData. ElementType is the data type of the created Value object. Currently, float and double are supported.

Parameters:

  • sampleShape: the tensor shape of the Value object.
  • batchData: the data to be contained in the Value object.
  • device: on which device the Value object should be created.
  • readOnly: the Value object is read-only if this flag is true.

template <typename ElementType> static ValuePtr CNTK::Value::CreateSequence(const NDShape& sampleShape, const std::vector<ElementType>& sequenceData, const DeviceDescriptor& device, bool readOnly = false)

Creates a new Value object containing a sequence of samples. The created Value object contains a copy of the specified data in sequenceData. The sequence length is the number of elements in sequenceData divided by the size of sampleShape (A runtime error occurs if the remainder is not zero). The created sequece is a new sequence. ElementType is the data type of the created Value object. Currently, float and double are supported.

Parameters:

  • sampleShape: the tensor shape of the Value.
  • sequenceData: the data to be contained in the Value.
  • device: on which device the Value object should be created.
  • readOnly: the Value is read-only if this flag is true.

template <typename ElementType> static ValuePtr CNTK::Value::CreateSequence(const NDShape& sampleShape, const std::vector<ElementType>& sequenceData, bool sequenceStartFlag, const DeviceDescriptor& device, bool readOnly = false)

Creates a new Value object containing a sequence of samples. The created Value object contains a copy of the specified data in sequenceData. The sequence length is the number of elements in sequenceData divided by the size of sampleShape (A runtime error occurs if the remainder is not zero). The sequenceStartFlag specifies whether this sequence is a new sequence or continuation of a previous sequence at the same index in the sequences list from a previous call to this method. ElementType is the data type of the created Value object. Currently, float and double are supported.

Parameters:

  • sampleShape: the tensor shape of the Value.
  • sequenceData: the data to be contained in the Value.
  • sequenceStartFlag: true indicates that it is a new sequence. false means a continuation of a previous sequence.
  • device: on which device the Value object should be created.
  • readOnly: the Value is read-only if this flag is true.

template <typename ElementType> static ValuePtr CNTK::Value::CreateBatchOfSequences(const NDShape& sampleShape, const std::vector<std::vector<ElementType>>& batchOfSequences, const DeviceDescriptor& device, bool readOnly = false)

Creates a new Value object containing a batch of variable length sequences. The created Value object contains a copy of the specified data in batchOfSequences. The number of sequences in the batch is the size of batchOfSequences. The length of each sequence is the number of elements in the corresponding sequence of batchOfSequences divided by the size of sampleShape (A runtime error occurs if the remainder is not zero). Each sequence in batchOfSequences is a new sequence. ElementType is the data type of the created Value object. Currently, float and double are supported.

Parameters:

  • sampleShape: the tensor shape of the Value.
  • batchOfSequences: the data to be stored in the Value. The outer vector represents a collection of sequences with variable length, and the inner vector represents each individual sequence.
  • device: on which device the Value should be created.
  • readOnly: the Value is read-only if this flag is true.

template <typename ElementType> static ValuePtr CNTK::Value::CreateBatchOfSequences(const NDShape& sampleShape, const std::vector<std::vector<ElementType>>& batchOfSequences, const std::vector<bool>& sequenceStartFlags, const DeviceDescriptor& device, bool readOnly = false)

Creates a new Value object containing a batch of variable length sequences. The created Value object contains a copy of the specified data in batchOfSequences. The number of sequences in the batch is the size of batchOfSequences. The length of each sequence is the number of elements in the corresponding sequence of batchOfSequences divided by the size of sampleShape (A runtime error occurs if the remainder is not zero). ElementType is the data type of the created Value object. Currently, float and double are supported.

Parameters:

  • sampleShape: the tensor shape of the Value.
  • batchOfSequences: the data to be stored in the Value. The outer vector represents a collection of sequences with variable length, and the inner vector represents each individual sequence.
  • sequenceStartFlags: A collection of boolean values. Each element represents whether the corresponding sequence in batchOfSequences is a new sequence (in case of true) or a continuation of a previous sequence (in case of false).
  • device: on which device the Value should be created.
  • readOnly: the Value is read-only if this flag is true.

The following methods create Value object from input data in one-hot vector format.

template <typename ElementType> static ValuePtr CNTK::Value::CreateBatch(size_t dimension, const std::vector<size_t>& batchData, const DeviceDescriptor& device, bool readOnly = false);

Creates a new Value object containing a batch of samples. Each sample is represented by an index value that points to the non-zero value in the one-hot vector of dimension elements. The number of samples in the batch is the number of elements in batchData. ElementType is the data type of the created Value object. Currently, float and double are supported.

Parameters:

  • dimension: the size of dimension of the one-hot vector.
  • batchData: the collection of indexes representing the batch of samples.
  • device: on which device the Value object should be created.
  • readOnly: the Value is read-only if this flag is true.

template <typename ElementType> static ValuePtr CNTK::Value::CreateSequence(size_t dimension, const std::vector<size_t>& sequenceData, const DeviceDescriptor& device, bool readOnly = false);

Creates a new Value object containing a sequence of samples. Each sample is represented by an index value that points to the non-zero value in the one-hot vector of dimension elements. The sequence length is the number of elements in sequenceData.Each sequence is a new sequence. ElementType is the data type of the created Value object. Currently, float and double are supported.

Parameters:

  • dimension: the size of dimension of the one-hot vector.
  • sequenceData: the collection of indexes representing the sequence of samples.
  • device: on which device the Value object should be created.
  • readOnly: the Value is read-only if this flag is true.

template <typename ElementType> static ValuePtr CNTK::Value::CreateSequence(size_t dimension, const std::vector<size_t>& sequenceData, bool sequenceStartFlag, const DeviceDescriptor& device, bool readOnly = false);

Creates a new Value object containing a sequence of samples. Each sample is represented by an index value that points to the non-zero value in the one-hot vector of dimension elements. The seqStartFlag specifies whether this sequence is a new sequence or continuation of a previous sequence at the same index in the sequences list from a previous call to this method. The sequence length is the number of elements in sequenceData. ElementType is the data type of the created Value object. Currently, float and double are supported.

Parameters:

  • dimension: the size of dimension of the one-hot vector.
  • sequenceData: the collection of indexes representing the sequence of samples.
  • sequenceStartFlag: true indicates that it is a new sequence. false means a continuation of a previous sequence.
  • device: on which device the Value object should be created.
  • readOnly: the Value is read-only if this flag is true.

template <typename ElementType> static ValuePtr CNTK::Value::CreateBatchOfSequences(size_t dimension, const std::vector<std::vector<size_t>>& batchOfSequences, const DeviceDescriptor& device, bool readOnly = false)

Creates a new Value object containing a batch of variable length sequences. Each sample is represented by an index value that points to the non-zero value in the one-hot vector of dimension elements. The number of sequences is the number of elements in the outer list of batchOfSequences. The length of each sequence is the number of elements of the corresponding sequence in the inner list of batchOfSequences. Each sequence in batchOfSequences is a new sequence. ElementType is the data type of the created Value object. Currently, float and double are supported.

Parameters:

  • dimension: the size of dimension of the one-hot vector.
  • batchOfSequences: the collection of indexes representing sequences of samples. The outer vector represents a collection of sequences with variable length, and the inner vector represents each individual sequence.
  • device: on which device the Value object should be created.
  • readOnly: the Value is read-only if this flag is true.

template <typename ElementType> static ValuePtr CNTK::Value::CreateBatchOfSequences(size_t dimension, const std::vector<std::vector<size_t>>& batchOfSequences, const std::vector<bool>& sequenceStartFlags, const DeviceDescriptor& device, bool readOnly = false)

Creates a new Value object containing a batch of variable length sequences. Each sample is represented by an index value that points to the non-zero value in the one-hot vector of dimension elements. The number of sequences is the number of elements in the outer list of batchOfSequences. The length of each sequence is the number of elements of the corresponding sequence in the inner list of batchOfSequences. ElementType is the data type of the created Value object. Currently, float and double are supported.

Parameters:

  • dimension: the size of dimension of the one-hot vector.
  • batchOfSequences: the collection of indexes representing sequences of samples. The outer vector represents a collection of sequences with variable length, and the inner vector represents each individual sequence.
  • sequenceStartFlags: A collection of boolean values. Each element represents whether the corresponding sequence in batchOfSequences is a new sequence (in case of true) or a continuation of a previous sequence (in case of false).
  • device: on which device the Value object should be created.
  • readOnly: the Value is read-only if this flag is true.

The following methods create Value object from input data in sparse CSC format.

Currently the Compressed Sparse Column Format (CSC) is supported. The CSC format stores the matrix in column-major format, and the array containing the column indices is compressed. A matrix in CSC format is represented by the following parameters:

  • nonZeroValues: the data array that holds all nonzero values of the matrix in column-major format.
  • rowIndices: the array that contains the row indices of the corresponding elements in array nonZeroValues.
  • colStarts: the array that holds indices into the arrays rowIndices and nonZeroValues.

A detailed description of the CSC format can be found here.

template <typename ElementType> static ValuePtr CNTK::Value::CreateSequence(const NDShape& sampleShape, size_t sequenceLength, const SparseIndexType* colStarts, const SparseIndexType* rowIndices, const ElementType* nonZeroValues, size_t numNonZeroValues, const DeviceDescriptor& device, bool readOnly = false)

Creates a new Value object containing a sequence of samples using CSC sparse input format. The sequence length is the number of rows of the sparse matrix. The created sequence is a new sequence. ElementType is the data type of the created Value object. Currently, float and double are supported.

Parameters:

  • sampleShape: the tensor shape of the Value. For sparse input, the tensor shape leading dimensionality must be the same as the total size of the tensor shape.
  • sequenceLength: the sequence length, which is also the number of rows in the sparse matrix.
  • colStarts: the array holds indices for each column into the arrays rowIndices and nonZeroValues.
  • rowIndices: the array that contains the row indices of the corresponding elements in array nonZeroValues.
  • nonZeroValues: the array that holds all nonzero values in the sparse matrix.
  • numNonZeroValues: the number of nonzero values in the sparse matrix.
  • device: on which device the Value object should be created.
  • readOnly: the Value is read-only if this flag is true.

template <typename ElementType> static ValuePtr CNTK::Value::CreateSequence(const NDShape& sampleShape, size_t sequenceLength, const SparseIndexType* colStarts, const SparseIndexType* rowIndices, const ElementType* nonZeroValues, size_t numNonZeroValues, bool sequenceStartFlag, const DeviceDescriptor& device, bool readOnly = false)

Creates a new Value object containing a sequence of samples using CSC sparse input format. The sequence length is the number of rows of the sparse matrix. The sequenceStartFlag specifies whether this sequence is a new sequence or continuation of a previous sequence from a previous call to this method. ElementType is the data type of the created Value object. Currently, float and double are supported.

Parameters:

  • sampleShape: the tensor shape of the Value. For sparse input, the tensor shape leading dimensionality must be the same as the total size of the tensor shape.
  • sequenceLength: the sequence length, which is also the number of rows in the sparse matrix.
  • colStarts: the array holds indices for each column into the arrays rowIndices and nonZeroValues.
  • rowIndices: the array that contains the row indices of the corresponding elements in array nonZeroValues.
  • nonZeroValues: the array that holds all nonzero values in the sparse matrix.
  • numNonZeroValues: the number of nonzero values in the sparse matrix.
  • sequenceStartFlag: true indicates that it is a new sequence. false means a continuation of a previous sequence.
  • device: on which device the Value object should be created.
  • readOnly: the Value is read-only if this flag is true.

template <typename ElementType> static ValuePtr CNTK::Value::CreateSequence(size_t dimension, size_t sequenceLength, const SparseIndexType* colStarts, const SparseIndexType* rowIndices, const ElementType* nonZeroValues, size_t numNonZeroValues, const DeviceDescriptor& device, bool readOnly = false)

Creates a new Value object containing a sequence of samples using CSC sparse input format. The sequence length is the number of rows of the sparse matrix. The created sequence is a new sequence. ElementType is the data type of the created Value object. Currently, float and double are supported.

Parameters:

  • dimension: the size of dimension of the one-hot vector.
  • sequenceLength: the sequence length, which is also the number of rows in the sparse matrix.
  • colStarts: the array holds indices for each column into the arrays rowIndices and nonZeroValues.
  • rowIndices: the array that contains the row indices of the corresponding elements in array nonZeroValues.
  • nonZeroValues: the array that holds all nonzero values in the sparse matrix.
  • numNonZeroValues: the number of nonzero values in the sparse matrix.
  • device: on which device the Value object should be created.
  • readOnly: the Value is read-only if this flag is true.

template <typename ElementType> static ValuePtr CNTK::Value::CreateSequence(size_t dimension, size_t sequenceLength, const SparseIndexType* colStarts, const SparseIndexType* rowIndices, const ElementType* nonZeroValues, size_t numNonZeroValues, bool sequenceStartFlag, const DeviceDescriptor& device, bool readOnly = false)

Creates a new Value object containing a sequence of samples using CSC sparse input format. The sequence length is the number of rows of the sparse matrix. The sequenceStartFlag specifies whether this sequence is a new sequence or continuation of a previous sequence. ElementType is the data type of the created Value object. Currently, float and double are supported.

Parameters:

  • dimension: the size of dimension of the one-hot vector.
  • sequenceLength: the sequence length, which is also the number of rows in the sparse matrix.
  • colStarts: the array holds indices for each column into the arrays rowIndices and nonZeroValues.
  • rowIndices: the array that contains the row indices of the corresponding elements in array nonZeroValues.
  • nonZeroValues: the array that holds all nonzero values in the sparse matrix.
  • numNonZeroValues: the number of nonzero values in the sparse matrix.
  • sequenceStartFlag: true indicates that it is a new sequence. false means a continuation of a previous sequence.
  • device: on which device the Value object should be created.
  • readOnly: the Value is read-only if this flag is true.

The following methods create a Value object from NDArrayView.

static ValuePtr CNTK::Value::Create(const NDShape& sampleShape, const std::vector<NDArrayViewPtr>& sequences, const DeviceDescriptor& device, bool readOnly = false)

Creates a new Value object based on a collection of NDArrayViews. Each sequence in sequences is a new sequence.

Parameters:

  • sampleShape: the tensor shape of the Value being created.
  • sequences: a collection of sequences represented by NDArrayView. Each NDArrayView represents a sequence.
  • device: on which device the Value object should be created.
  • readOnly: the Value is read-only if this flag is true.

static ValuePtr CNTK::Value::Create(const NDShape& sampleShape, const std::vector<NDArrayViewPtr>& sequences, const std::vector<bool>& sequenceStartFlags, const DeviceDescriptor& device, bool readOnly, bool createNewCopy)

Creates a new Value object based on a collection of NDArrayViews. The sequenceStartFlags specifies whether a sequence is a new sequence or continuation of a previous sequence.

Parameters:

  • sampleShape: the tensor shape of the Value being created.
  • sequences: a collection of sequences represented by NDArrayView. Each NDArrayView represents a sequence.
  • sequenceStartFlags: A collection of boolean values. Each element represents whether the corresponding sequence in sequences is a new sequence (in case of true) or a continuation of a previous sequence (in case of false).
  • device: on which device the Value object should be created.
  • readOnly: the Value is read-only if this flag is true.

The following methods copy data stored in a Value object as dense or one-hot vector output.

template <typename ElementType> void CNTK::Value::CopyVariableValueTo(const Variable& outputVariable, std::vector<std::vector<ElementType>>& sequences)

Copies the data stored in the Value into the buffer provided by sequences. The sequences is a list of sequences with variable length. The number of items contained in the outer list of sequences is the number of sequences in the Value. Each element of the outer list represents a sequence. Each sequence, represented by vector<ElementType>, contains a variable number of samples. Each sample consits of a fixed number of elements with type of ElementType. The number of elements of a sample is determined by the shape of outputVariable. The shape of the variable should match the shape of the Value. ElementType is the data type of the created Value object. Currently, float and double are supported.

Parameters:

  • outputVariable: denotes the shape and dynamic axes when copying data from this Value to the sequences.
  • sequences: the output buffer used to store the data copied from the Value.

void CNTK::Value::CopyVariableValueTo(const Variable& outputVariable, std::vector<std::vector<size_t>>& sequences)

Copies the data stored in the Value object into the buffer provided by sequences. The sequences is a list of sequences with variable length. The number of items contained in the outer list of sequences is the number of sequences in the Value. Each element of the outer list represents a sequence. Each sequence, represented by vector<size_t>, contains a variable number of samples. Each sample is represented by an index pointing to the non-zero value in the one-hot vector. The dimension size of the one-hot vector should match that defined in the outputVariable.

Parameters:

  • outputVariable: denotes the shape and dynamic axes when copying data from this Value to the sequences.
  • sequences: the output buffer used to store the data copied from the Value.

Please refer to CNTKLibrary.h for details of data types used by the functions above.

Examples

The C++ examples CNTKLibraryCPPEvalCPUOnlyExamples and CNTKLibraryCPPEvalGPUExamples demonstrate the usage of CNTK Library for evaluation. The samples also show how to do multiple evaluations in parallel using multiple threads and share model parameters among threads.

New Documentation Site

Iteration Plans

Clone this wiki locally