seldonian.RL.Agents.Policies.Policy.Policy¶
- class Policy¶
Bases:
object
- __init__()¶
Base class for policies. Contains methods, some of which must be overridden in any policy implementation
- __repr__()¶
Return repr(self).
Methods
- choose_action(obs)¶
Defines how to select an action given an observation, obs
- get_params()¶
Get current policy parameters
- get_prob_this_action(obs, action)¶
Get probability of taking an action given an observation. Does not necessarily need to be overridden, but is often called from self.get_probs_from_observations_and_actions()
- get_probs_from_observations_and_actions(observations, actions, behavior_action_probs)¶
Get probabilities for each observation and action in the input arrays
- set_new_params(new_params)¶
Update policy parameters