Operator structure. More...

#include <op.h>

Collaboration diagram for nnvm::Op:

Public Member Functions
Op &	describe (const std::string &descr)
	setter function during registration Set the description of operator More...

Op &	add_argument (const std::string &name, const std::string &type, const std::string &description)
	Add argument information to the function. More...

Op &	add_arguments (const std::vector< ParamFieldInfo > &args)
	Append list if arguments to the end. More...

Op &	set_num_inputs (uint32_t n)
	Set the num_inputs. More...

Op &	set_support_level (uint32_t level)
	Set the support level of op. More...

Op &	set_num_inputs (std::function< uint32_t(const NodeAttrs &attr)> fn)
	Set the get_num_outputs function. More...

Op &	set_num_outputs (uint32_t n)
	Set the num_outputs. More...

Op &	set_num_outputs (std::function< uint32_t(const NodeAttrs &attr)> fn)
	Set the get_num_outputs function. More...

Op &	set_attr_parser (std::function< void(NodeAttrs *attrs)> fn)
	Set the attr_parser function. More...

template<typename ValueType >
Op &	set_attr (const std::string &attr_name, const ValueType &value, int plevel=10)
	Register additional attributes to operator. More...

Op &	add_alias (const std::string &alias)
	Add another alias to this operator. The same Op can be queried with Op::Get(alias) More...

Op &	include (const std::string &group_name)
	Include all the attributes from an registered op group. More...

Static Public Member Functions
static const Op *	Get (const std::string &op_name)
	Get an Op for a given operator name. Will raise an error if the op has not been registered. More...

template<typename ValueType >
static const OpMap< ValueType > &	GetAttr (const std::string &attr_name)
	Get additional registered attribute about operators. If nothing has been registered, an empty OpMap will be returned. More...

Public Attributes
std::string	name
	name of the operator More...

std::string	description
	detailed description of the operator This can be used to generate docstring automatically for the operator. More...

std::vector< ParamFieldInfo >	arguments

uint32_t	num_inputs = 1
	number of inputs to the operator, -1 means it is variable length When get_num_inputs is presented, the number will be decided by get_num_inputs instead. More...

uint32_t	num_outputs = 1
	number of outputs of the operator When get_num_outputs is presented. The number of outputs will be decided by get_num_outputs function More...

uint32_t	support_level = 10
	support level of the operator, The lower the more priority it contains. This is in analogies to BLAS levels. More...

std::function< uint32_t(const NodeAttrs &attrs)>	get_num_outputs = nullptr
	get number of outputs given information about the node. More...

std::function< uint32_t(const NodeAttrs &attrs)>	get_num_inputs = nullptr
	get number of inputs given information about the node. More...

std::function< void(NodeAttrs *attrs)>	attr_parser = nullptr
	Attribute parser to parse the NodeAttrs information. More...

Friends
template<typename ValueType >
class	OpMap

class	OpGroup

class	dmlc::Registry< Op >

Detailed Description

Operator structure.

Besides the fields in the structure, arbitary additional information can be associated with each op. See function GetAttr for details.

 // Example usage of Op
 // registeration of oeprators
 // NOTE that the attr function can register any
 // additional attributes to the operator
 NNVM_REGISTER_OP(add)
 .describe("add two inputs together")
 .set_num_inputs(2)
 .set_attr<OpKernel>("OpKernel<gpu>", AddKernel)
 .include("ElementwiseOpAttr");
 // can register attribute by group
 // all the ops that include the group get the attribute.
 NNVM_REGISTER_OP_GROUP(ElementwiseOpAttr)
 .set_attr<FInferShape>("FInferShape", ElementwiseInferShape);
 NNVM_REGISTER_OP(sub)
 .describe("substract one tensor from another")
 .set_num_inputs(2);
 // Can call regster multiple times in different files
 // to register different part of information
 NNVM_REGISTER_OP(sub)
 .set_attr<OpKernel>("OpKernel<gpu>", SubKernel);
 .include("ElementwiseOpAttr");
 // get operators from registry.
 void my_function() {
   const Op* add = Op::Get("add");
   const Op* sub = Op::Get("sub");
   // query basic information about each operator.
   assert(op->name == "plus");
   assert(op->num_inputs == 2);
   // get additional registered information,
   // Assume user registered a OpKernel type attribute as gpu_kernel on each operator.
   const OpMap<OpKernel>& kernel = Op::GetAttr<OpKernel>("OpKernel<gpu>");
   // we can get the kernel functions by using operator as key.
   auto add_kernel = kernel[add];
   auto sub_kernel = kernel[sub];
   // subsequent code can make use of the queried kernel functions.
}

Member Function Documentation

◆ add_alias()

Op& nnvm::Op::add_alias ( const std::string & alias )

Add another alias to this operator. The same Op can be queried with Op::Get(alias)

Parameters

alias The alias of the operator.

Returns: reference to self.

◆ add_argument()

Op & nnvm::Op::add_argument	(	const std::string &	name,
		const std::string &	type,
		const std::string &	description
	)

inline

Add argument information to the function.

Parameters

name	Name of the argument.
type	Type of the argument.
description	Description of the argument.

Returns: reference to self.

◆ add_arguments()

Op & nnvm::Op::add_arguments ( const std::vector< ParamFieldInfo > & args )

inline

Append list if arguments to the end.

Parameters

args	Additional list of arguments.

Returns: reference to self.

◆ describe()

Op & nnvm::Op::describe ( const std::string & descr )

inline

setter function during registration Set the description of operator

Parameters

descr the description string.

Returns: reference to self.

◆ Get()

static const Op* nnvm::Op::Get ( const std::string & op_name )

static

Get an Op for a given operator name. Will raise an error if the op has not been registered.

Parameters

op_name Name of the operator.

Returns: Pointer to a Op, valid throughout program lifetime.

◆ GetAttr()

template<typename ValueType >

const OpMap< ValueType > & nnvm::Op::GetAttr ( const std::string & attr_name )

inlinestatic

Get additional registered attribute about operators. If nothing has been registered, an empty OpMap will be returned.

Parameters

attr_name The name of the attribute.

Returns: An OpMap of specified attr_name.

Template Parameters

ValueType The type of the attribute.

◆ include()

Op& nnvm::Op::include ( const std::string & group_name )

Include all the attributes from an registered op group.

Parameters

group_name The name of the group.

Returns: reference to self.

See also: NNVM_REGISTER_OP_GROUP

◆ set_attr()

template<typename ValueType >

Op & nnvm::Op::set_attr	(	const std::string &	attr_name,
		const ValueType &	value,
		int	plevel = `10`
	)

inline

Register additional attributes to operator.

Parameters

attr_name	The name of the attribute.
value	The value to be set.
plevel	The priority level of this set, an higher priority level attribute will replace lower priority level attribute. Must be bigger than 0.

Cannot set with same plevel twice in the code.

Template Parameters

ValueType The type of the value to be set.

◆ set_attr_parser()

Op & nnvm::Op::set_attr_parser ( std::function< void(NodeAttrs *attrs)> fn )

inline

Set the attr_parser function.

Parameters

fn	The number of outputs to be set.

Returns: reference to self.

◆ set_num_inputs() [1/2]

Op & nnvm::Op::set_num_inputs ( uint32_t n )

inline

Set the num_inputs.

Parameters

n	The number of inputs to be set.

Returns: reference to self.

◆ set_num_inputs() [2/2]

Op & nnvm::Op::set_num_inputs ( std::function< uint32_t(const NodeAttrs &attr)> fn )

inline

Set the get_num_outputs function.

Parameters

fn	The function to be set.

Returns: reference to self.

◆ set_num_outputs() [1/2]

Op & nnvm::Op::set_num_outputs ( uint32_t n )

inline

Set the num_outputs.

Parameters

n	The number of outputs to be set.

Returns: reference to self.

◆ set_num_outputs() [2/2]

Op & nnvm::Op::set_num_outputs ( std::function< uint32_t(const NodeAttrs &attr)> fn )

inline

Set the get_num_outputs function.

Parameters

fn	The function to be set.

Returns: reference to self.

◆ set_support_level()

Op & nnvm::Op::set_support_level ( uint32_t level )

inline

Set the support level of op.

Parameters

level The support level.

Returns: reference to self.

Friends And Related Function Documentation

◆ dmlc::Registry< Op >

friend class dmlc::Registry< Op >

friend

◆ OpGroup

friend class OpGroup

friend

◆ OpMap

template<typename ValueType >

friend class OpMap

friend

Member Data Documentation

◆ arguments

std::vector<ParamFieldInfo> nnvm::Op::arguments

◆ attr_parser

std::function<void(NodeAttrs* attrs)> nnvm::Op::attr_parser = nullptr

Attribute parser to parse the NodeAttrs information.

This can help to get quick access to a parsed attribute object

// Example usage of attr_parser.
// Suppose we want to register operator sum.
// The parameters about sum operator
struct SumParam {
  int axis;
};
// The parser function
void SumAttrParser(NodeAttrs* attrs) {
   // This will be invoked during node construction.
   SumParam param;
   // parse axis string to integer
   param.axis = atoi(attrs->dict["axis"].c_str());
   // set the parsed parameter
   attrs->parsed = std::move(param);
}
// The other function that can utilize the parsed result.
TShape SumInferShape(const NodeAttrs& attrs,
                     const std::vector<TShape>& ishapes) {
   // we can use the parsed version of param
   // without repeatively parsing the parameter
   const SumParam& param = nnvm::get<SumParam>(attrs.parsed);
}

◆ description

std::string nnvm::Op::description

detailed description of the operator This can be used to generate docstring automatically for the operator.

◆ get_num_inputs

std::function<uint32_t(const NodeAttrs& attrs)> nnvm::Op::get_num_inputs = nullptr

get number of inputs given information about the node.

Parameters

attrs The attribute of the node

Returns: number of inputs

◆ get_num_outputs

std::function<uint32_t(const NodeAttrs& attrs)> nnvm::Op::get_num_outputs = nullptr

get number of outputs given information about the node.

Parameters

attrs The attribute of the node

Returns: number of outputs.

◆ name

std::string nnvm::Op::name

name of the operator

◆ num_inputs

uint32_t nnvm::Op::num_inputs = 1

number of inputs to the operator, -1 means it is variable length When get_num_inputs is presented, the number will be decided by get_num_inputs instead.

See also: get_num_inputs

◆ num_outputs

uint32_t nnvm::Op::num_outputs = 1

number of outputs of the operator When get_num_outputs is presented. The number of outputs will be decided by get_num_outputs function

See also: get_num_outputs

◆ support_level

uint32_t nnvm::Op::support_level = 10

support level of the operator, The lower the more priority it contains. This is in analogies to BLAS levels.

The documentation for this class was generated from the following file:

/work/mxnet/3rdparty/tvm/nnvm/include/nnvm/op.h

Public Member Functions

Static Public Member Functions

Public Attributes

Friends

Detailed Description

Member Function Documentation

◆ add_alias()

◆ add_argument()

◆ add_arguments()

◆ describe()

◆ Get()

◆ GetAttr()

◆ include()

◆ set_attr()

◆ set_attr_parser()

◆ set_num_inputs() [1/2]

◆ set_num_inputs() [2/2]

◆ set_num_outputs() [1/2]

◆ set_num_outputs() [2/2]

◆ set_support_level()

Friends And Related Function Documentation

◆ dmlc::Registry< Op >

◆ OpGroup

◆ OpMap

Member Data Documentation

◆ arguments

◆ attr_parser

◆ description

◆ get_num_inputs

◆ get_num_outputs

◆ name

◆ num_inputs

◆ num_outputs

◆ support_level