src.gridmind.algorithms.base_learning_algorithm

Attributes

SAVE_DATA_DIR

Classes

BaseLearningAlgorithm

Helper class that provides a standard way to create an ABC using

Module Contents

src.gridmind.algorithms.base_learning_algorithm.SAVE_DATA_DIR = None

class src.gridmind.algorithms.base_learning_algorithm.BaseLearningAlgorithm(name: str, env: gymnasium.Env | None = None, summary_dir: str | None = None, write_summary: bool = True)

Bases: abc.ABC

Helper class that provides a standard way to create an ABC using inheritance.

name

logger = None

env = None

epoch_eval_interval = None

perform_evaluation = False

monitor_divergence = False

stop_on_divergence = False

write_summary = True

_initialize_summary_writer(summary_dir, env_name, extra_info: str = '', use_async_writer: bool = False)

register_performance_evaluator(evaluator: gridmind.utils.performance_evaluation.base_performance_evaluator.BasePerformanceEvaluator)

register_divergence_detector(detector: gridmind.utils.divergence.base_divergence_detector.BaseDivergenceDetector)

report_policy()

report_state_values()

report_state_action_values()

_preprocess(observation)

speculate_divergence()

abstract _get_state_value_fn(force_functional_interface: bool = True)

abstract _get_state_action_value_fn(force_functional_interface: bool = True)

abstract _get_policy()

get_state_value_fn(force_functional_interface: bool = True, autopreprocess: bool = False)

get_state_action_value_fn(force_functional_interface: bool = True, autopreprocess: bool = False)

get_policy(autopreprocess: bool = False)

abstract set_policy(policy: gridmind.policies.base_policy.BasePolicy, **kwargs)

abstract _train_episodes(num_episodes: int, prediction_only: bool, *args, **kwargs)

get_policy_cloned()

train(num_episodes: int | None = None, num_steps: int | None = None, prediction_only: bool = False, save_policy: bool = True, *args, **kwargs)

train_steps(num_steps: int, prediction_only: bool, save_policy: bool = True, *args, **kwargs)

abstract _train_steps(num_steps: int, prediction_only: bool, *args, **kwargs)

train_episodes(num_episodes: int, prediction_only: bool, save_policy: bool = True, *args, **kwargs)

_training_wrapper(num_iter: int, prediction_only: bool, save_policy: bool, training_fn: Callable, *args, **kwargs)

_report_all_metrics()

evaluate_policy(num_episodes: int)

optimize_policy(num_episodes: int)

save_policy(path: str)

static load_policy(saved_policy_path: str)