choose_rank() returns the action choice using ranking_response. ranking_response contains all the actions and distribution from with the action was sampled. It also contains the chosen action id and the unique event_id representing the choice. This unique event_id must be used to report back outcomes against this choice for the online trainer. More...

#include <ranking_response.h>

Public Types
using	iterator = slot_ranking::iterator

using	const_iterator = slot_ranking::const_iterator

Public Member Functions
	ranking_response (char const *event_id)
	Construct a new ranking response object. More...

const char *	get_event_id () const
	Unique event_id for this ranking request. This event_id must be used when calling report_outcome so it can be joined with the chosen action. More...

void	set_event_id (const char *event_id)
	Set the event_id. (This is set internally by the API) More...

int	get_chosen_action_id (size_t &action_id, api_status *status=nullptr) const
	Get the chosen action id. More...

int	set_chosen_action_id (size_t action_id, api_status *status=nullptr)
	Set the chosen action id. (This is set internally by the API) More...

int	set_chosen_action_id_unchecked (size_t action_id, api_status *status=nullptr)
	Set the chosen action id, but do not verify the index fits within the ranking. (This is set internally by the API) This is used in CCB where subsequent ranking_responses have subsets of the orignal actionset. More...

void	push_back (const size_t action_id, const float prob)
	Add (action id, probability) pair to the slot (This is set internally by the API) More...

size_t	size () const
	Size of the action collection. More...

void	set_model_id (const char *model_id)
	Set the model_id. Every rank call is associated with a unique model used to predict. A unique model_id is associated with each unique model. (This is set internally by the API) More...

void	set_model_id (std::string &&model_id)
	Set the model_id. Every rank call is associated with a unique model used to predict. A unique model_id is associated with each unique model. (This is set internally by the API). Input model_id is left in an unspecified but valid state. More...

const char *	get_model_id () const
	Get the model_id. Every rank call (single or multi slot) is associated with a unique model used to predict. A unique model_id is associated with each unique model. (This is set internally by the API) More...

void	clear ()
	Clear the ranking response object so that it can be reused. The goal is to reuse response without reallocating as much as possible.

	ranking_response (ranking_response &&) noexcept
	Move construct a new ranking response object. The underlying data is taken from the rvalue reference.

ranking_response &	operator= (ranking_response &&) noexcept
	Move assignment operator for ranking response. The underlying data is taken from rvalue reference, and then it is cleared. More...

	ranking_response (const ranking_response &)=delete
	Copy constructor is removed since implementation will be deleted twice.

ranking_response &	operator= (const ranking_response &)=delete
	assignment operator is removed since implementation will be deleted twice

const_iterator	begin () const
	Returns an iterator pointing to the first element of the (action, probability) collection.

iterator	begin ()

const_iterator	end () const
	Returns an iterator referring to the past-the-end element of the (action, probability) collection.

iterator	end ()

Detailed Description

choose_rank() returns the action choice using ranking_response. ranking_response contains all the actions and distribution from with the action was sampled. It also contains the chosen action id and the unique event_id representing the choice. This unique event_id must be used to report back outcomes against this choice for the online trainer.

Examples: rl_sim.cc.

Constructor & Destructor Documentation

◆ ranking_response()

reinforcement_learning::ranking_response::ranking_response ( char const * event_id )

Construct a new ranking response object.

Parameters

event_id The unique identifier for this interaction. This event_id must be used when reporting the outcome for this action

Member Function Documentation

◆ get_chosen_action_id()

int reinforcement_learning::ranking_response::get_chosen_action_id	(	size_t &	action_id,
		api_status *	status = `nullptr`
	)		const

Get the chosen action id.

Parameters

action_id	Chosen action id
status	Optional field with detailed string description if there is an error

Returns: int Error code

◆ get_event_id()

const char* reinforcement_learning::ranking_response::get_event_id ( ) const

Unique event_id for this ranking request. This event_id must be used when calling report_outcome so it can be joined with the chosen action.

Returns: const char*

◆ get_model_id()

const char* reinforcement_learning::ranking_response::get_model_id ( ) const

Get the model_id. Every rank call (single or multi slot) is associated with a unique model used to predict. A unique model_id is associated with each unique model. (This is set internally by the API)

Returns: const char*

◆ operator=()

ranking_response& reinforcement_learning::ranking_response::operator= ( ranking_response && )

noexcept

Move assignment operator for ranking response. The underlying data is taken from rvalue reference, and then it is cleared.

Returns: ranking_response&

◆ push_back()

void reinforcement_learning::ranking_response::push_back	(	const size_t	action_id,
		const float	prob
	)

Add (action id, probability) pair to the slot (This is set internally by the API)

Parameters

action_id
prob

◆ set_chosen_action_id()

int reinforcement_learning::ranking_response::set_chosen_action_id	(	size_t	action_id,
		api_status *	status = `nullptr`
	)

Set the chosen action id. (This is set internally by the API)

Parameters

action_id	Chosen action id
status	Optional field with detailed string description if there is an error

Returns: int Error code

◆ set_chosen_action_id_unchecked()

int reinforcement_learning::ranking_response::set_chosen_action_id_unchecked	(	size_t	action_id,
		api_status *	status = `nullptr`
	)

Set the chosen action id, but do not verify the index fits within the ranking. (This is set internally by the API) This is used in CCB where subsequent ranking_responses have subsets of the orignal actionset.

Parameters

action_id	Chosen action id
status	Optional field with detailed string description if there is an error

Returns: int Error code

◆ set_event_id()

void reinforcement_learning::ranking_response::set_event_id ( const char * event_id )

Set the event_id. (This is set internally by the API)

Parameters

event_id

◆ set_model_id() [1/2]

void reinforcement_learning::ranking_response::set_model_id ( const char * model_id )

Set the model_id. Every rank call is associated with a unique model used to predict. A unique model_id is associated with each unique model. (This is set internally by the API)

Parameters

model_id

◆ set_model_id() [2/2]

void reinforcement_learning::ranking_response::set_model_id ( std::string && model_id )

Set the model_id. Every rank call is associated with a unique model used to predict. A unique model_id is associated with each unique model. (This is set internally by the API). Input model_id is left in an unspecified but valid state.

Parameters

model_id

◆ size()

size_t reinforcement_learning::ranking_response::size ( ) const

Size of the action collection.

Returns: size_t

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

/home/runner/work/reinforcement_learning/reinforcement_learning/include/ranking_response.h

Public Types

Public Member Functions

Detailed Description

Constructor & Destructor Documentation

◆ ranking_response()

Member Function Documentation

◆ get_chosen_action_id()

◆ get_event_id()

◆ get_model_id()

◆ operator=()

◆ push_back()

◆ set_chosen_action_id()

◆ set_chosen_action_id_unchecked()

◆ set_event_id()

◆ set_model_id() [1/2]

◆ set_model_id() [2/2]

◆ size()