ngraph::op Namespace Reference

Ops used in graph-building. More...

Data Structures
struct	AutoBroadcastSpec
	Implicit broadcast specification. More...
struct	BroadcastModeSpec
	Implicit broadcast specification. More...
struct	DetectionOutputAttrs
class	Op
	Root of all actual ops. More...
struct	PriorBoxAttrs
struct	PriorBoxClusteredAttrs
struct	ProposalAttrs

Enumerations
enum	LSTMWeightsFormat {   FICO, ICOF, IFCO, IFOC,   IOFC }
enum	PadMode { CONSTANT = 0, EDGE, REFLECT, SYMMETRIC }
	Modes for the Pad operator.
enum	PadType {   EXPLICIT = 0, SAME_LOWER, SAME_UPPER, VALID,   AUTO = SAME_UPPER, NOTSET = EXPLICIT }
	Padding Type used for Convolution and Pooling More...
enum	RoundingType { FLOOR = 0, CEIL = 1 }
	Rounding Type used for Pooling operators.
enum	AutoBroadcastType { NONE = 0, EXPLICIT = NONE, NUMPY, PDPD }
	Specifies the algorithm to use for implicit broadcasting of a tensor to align with another tensor. More...
enum	BroadcastType {   NONE, EXPLICIT = NONE, NUMPY, PDPD,   BIDIRECTIONAL }
	BroadcastType specifies rules used for mapping of input tensor axes to output shape axes. More...
enum	EpsMode { ADD, MAX }
	Specifies how eps is combined with L2 value.
enum	TopKSortType { NONE, SORT_INDICES, SORT_VALUES }
enum	TopKMode { MAX, MIN }
enum	RecurrentSequenceDirection { FORWARD, REVERSE, BIDIRECTIONAL }
	This class defines possible recurrent sequence directions.

Functions
NGRAPH_API std::ostream &	operator<< (std::ostream &s, const PadMode &type)
NGRAPH_API std::ostream &	operator<< (std::ostream &s, const PadType &type)
NGRAPH_API std::ostream &	operator<< (std::ostream &s, const RoundingType &type)
NGRAPH_API std::ostream &	operator<< (std::ostream &s, const AutoBroadcastType &type)
NGRAPH_API std::ostream &	operator<< (std::ostream &s, const BroadcastType &type)
NGRAPH_API std::ostream &	operator<< (std::ostream &s, const EpsMode &type)
NGRAPH_API std::ostream &	operator<< (std::ostream &s, const TopKSortType &type)
NGRAPH_API std::ostream &	operator<< (std::ostream &s, const TopKMode &type)
NGRAPH_API std::ostream &	operator<< (std::ostream &s, const RecurrentSequenceDirection &direction)
NGRAPH_API bool	is_unary_elementwise_arithmetic (const ngraph::Node *node)
NGRAPH_API bool	is_binary_elementwise_arithmetic (const ngraph::Node *node)
NGRAPH_API bool	is_binary_elementwise_comparison (const ngraph::Node *node)
NGRAPH_API bool	is_binary_elementwise_logical (const ngraph::Node *node)
NGRAPH_API bool	supports_auto_broadcast (const ngraph::Node *node)
NGRAPH_API bool	supports_decompose (const ngraph::Node *node)
NGRAPH_API bool	is_op (const ngraph::Node *node)
NGRAPH_API bool	is_parameter (const ngraph::Node *node)
NGRAPH_API bool	is_output (const ngraph::Node *node)
NGRAPH_API bool	is_constant (const ngraph::Node *node)
NGRAPH_API bool	is_commutative (const ngraph::Node *node)
NGRAPH_API bool	is_unary_elementwise_arithmetic (const std::shared_ptr< ngraph::Node > &node)
NGRAPH_API bool	is_binary_elementwise_arithmetic (const std::shared_ptr< ngraph::Node > &node)
NGRAPH_API bool	is_binary_elementwise_comparison (const std::shared_ptr< ngraph::Node > &node)
NGRAPH_API bool	is_binary_elementwise_logical (const std::shared_ptr< ngraph::Node > &node)
NGRAPH_API bool	supports_auto_broadcast (const std::shared_ptr< ngraph::Node > &node)
NGRAPH_API bool	supports_decompose (const std::shared_ptr< ngraph::Node > &node)
NGRAPH_API bool	is_op (const std::shared_ptr< ngraph::Node > &node)
NGRAPH_API bool	is_parameter (const std::shared_ptr< ngraph::Node > &node)
NGRAPH_API bool	is_output (const std::shared_ptr< ngraph::Node > &node)
NGRAPH_API bool	is_constant (const std::shared_ptr< ngraph::Node > &node)
NGRAPH_API bool	is_commutative (const std::shared_ptr< ngraph::Node > &node)
ONNX_IMPORTER_API bool	is_null (const ngraph::Node *node)
ONNX_IMPORTER_API bool	is_null (const std::shared_ptr< ngraph::Node > &node)
ONNX_IMPORTER_API bool	is_null (const Output< ngraph::Node > &output)

Detailed Description

Ops used in graph-building.

Enumeration Type Documentation

AutoBroadcastType

enum ngraph::op::AutoBroadcastType

strong

Specifies the algorithm to use for implicit broadcasting of a tensor to align with another tensor.

NONE - No implicit broadcasting of tensor NUMPY - Numpy-style implicit broadcasting (https://docs.scipy.org/doc/numpy/user/basics.broadcasting.html) Right-align dimensions of the two tensors, with missing dimensions treated as size 1 dimensions. After alignment, for each dimension, their sizes should either match or one of them should be of size 1. Size 1 dimension will be implicitly broadcast to match the other size.

E.g., A: Shape(2, 1, 6) B: Shape( 3, 1) Result: Shape(2, 3, 6)

 A: Shape(2, 1, 6)
 B: Shape(   3, 1)

Result: Shape(2, 3, 6) PDPD - PaddlePaddle-style implicit broadcasting (https://github.com/PaddlePaddle/Paddle/blob/release/1.5/paddle/ fluid/operators/elementwise/elementwise_op.h#L126) Broadcast B to match the shape of A, where axis is the start dimension index to align B with A. If axis is -1 (default), i axis = rank(A) - rank(B). The trailing dimensions of size 1 for B will be ignored.

E.g., A: Shape(2, 3, 4, 5) B: Shape( 3, 4 ) with axis =1 Result: Shape(2, 3, 4, 5)

 A: Shape(2, 3, 4, 5)
 B: Shape(   3, 1   ) with axis = 1

Result: Shape(2, 3, 4, 5)

TODO: Add more implicit broadcast modes used by frameworks

BroadcastType

enum ngraph::op::BroadcastType

strong

BroadcastType specifies rules used for mapping of input tensor axes to output shape axes.

Note

Broadcasting rules are different for Broadcast op and for element-wise ops. AutoBroadcastType::NUMPY is equivalent of BroadcastType::BIDIRECTIONAL according to spec.

EXPLICIT - Mapping of the input data shape to output shape based on axes_mapping input. NUMPY - Numpy broadcasting rules, aligned with ONNX Broadcasting. (https://github.com/onnx/onnx/blob/master/docs/Broadcasting.md) PDPD - PaddlePaddle-style implicit broadcasting. For more informaction see AutoBroadcastType documentation. BIDIRECTIONAL - The broadcast rule is similar to numpy.array(input) * numpy.ones(target_shape). Dimensions are right alignment.

PadType

enum ngraph::op::PadType

strong

Padding Type used for Convolution and Pooling

Follows ONNX padding type definitions EXPLICIT - Pad dimensions are explicity specified SAME_LOWER - Pad dimensions computed to match input shape Ceil(num_dims/2) at the beginning and Floor(num_dims/2) at the end SAME_UPPER - Pad dimensions computed to match input shape Floor(num_dims/2) at the beginning and Ceil(num_dims/2) at the end VALID - No padding