sarsa
=====

.. py:module:: sarsa


Classes
-------

.. autoapisummary::

   sarsa.SARSA


Module Contents
---------------

.. py:class:: SARSA(env: gymnasium.Env, policy: Optional[gridmind.policies.soft.q_derived.base_q_derived_soft_policy.BaseQDerivedSoftPolicy] = None, step_size: float = 0.5, discount_factor: float = 0.9, q_initializer: str = 'zero', epsilon_decay: bool = False, feature_constructor: Callable = None, summary_dir: Optional[str] = None, write_summary: bool = True)

   Bases: :py:obj:`gridmind.algorithms.base_learning_algorithm.BaseLearningAlgorithm`


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


   .. py:attribute:: num_actions


   .. py:attribute:: feature_constructor
      :value: None



   .. py:attribute:: policy
      :value: None



   .. py:attribute:: step_size
      :value: 0.5



   .. py:attribute:: discount_factor
      :value: 0.9



   .. py:attribute:: epsilon_decay
      :value: False



   .. py:method:: _get_state_value_fn(force_functional_interface: bool = True)


   .. py:method:: _get_state_action_value_fn(force_functional_interface: bool = True)


   .. py:method:: _get_policy()


   .. py:method:: _train_steps(num_steps: int, prediction_only: bool, *args, **kwargs)
      :abstractmethod:



   .. py:method:: _train_episodes(num_episodes: int, prediction_only: bool = False)


   .. py:method:: set_policy(policy: gridmind.policies.soft.q_derived.base_q_derived_soft_policy.BaseQDerivedSoftPolicy)