cuberl::rl Namespace Reference

Namespaces
namespace	agents
namespace	algos
namespace	envs
namespace	policies

Classes
struct	ConstantLRScheduler
	struct ConstantLRScheduler. Echos bach the learning rate it was given More...
struct	EpisodeInfo
	The EpisodeInfo struct. More...
class	RLSerialAgentTrainer
struct	RLSerialTrainerConfig
	The RLSerialTrainerConfig struct. Configuration struct for the serial RL agent trainer. More...
struct	with_decay_epsilon_option_mixin
	The with_decay_epsilon_option_mixin struct. More...
struct	with_double_q_table_max_action_mixin
struct	with_double_q_table_mixin
struct	with_double_q_table_mixin< DynMat< real_t > >
	The WithDoubleQTableMixin struct. More...
struct	with_double_q_table_mixin< std::map< KeyTp, DynVec< real_t > > >
struct	with_q_table_mixin
	The WithQTableMixin struct. More...

Enumerations
enum class	EpsilonDecayOptionType : int {   INVALID_TYPE =0 , NONE , EXPONENTIAL , INVERSE_STEP ,   CONSTANT_RATE }

Functions
std::ostream &	operator<< (std::ostream &out, const EpisodeInfo &info)
uint_t	max_action (const DynMat< real_t > &qtable, uint_t state, uint_t n_actions)
	max_action

Detailed Description

Utility class used in the implementation of RL algorithms

Enumeration Type Documentation

EpsilonDecayOptionType

enum class cuberl::rl::EpsilonDecayOptionType : int

strong

Enumerator
INVALID_TYPE
NONE
EXPONENTIAL
INVERSE_STEP
CONSTANT_RATE

Function Documentation

max_action()

uint_t cuberl::rl::max_action	(	const DynMat< real_t > &	qtable,
		uint_t	state,
		uint_t	n_actions
	)

max_action

Parameters

qtable
state
n_actions

Returns

operator<<()

std::ostream & cuberl::rl::operator<< ( std::ostream &  out, const EpisodeInfo &  info  )

inline