def __init__(self, learner, sdim, adim=1, maxHistoryLength=1000, batch=False): LoggingAgent.__init__(self, sdim, adim) self.learner = learner self.policy = self.learner.module.policy self.lastaction = None self.learning = True self.batch = batch self.currentDataIndex = 0 self.maxHistoryLength = maxHistoryLength
def __init__(self, learner, **kwargs): LoggingAgent.__init__(self,learner.num_features,learner.num_actions, **kwargs) self.learner = learner self.reset() self.learning=True self.learner.dataset=self.history self.visited_states_x=[] self.visited_states_y=[] self.qvalues=[] self.actionvalues=[] self.init_exploration=1.0
def __init__(self, module, learner = None): """ :key module: the acting module :key learner: the learner (optional) """ LoggingAgent.__init__(self, module.indim, module.outdim) self.module = module self.learner = learner # if learner is available, tell it the module and data if self.learner is not None: self.learner.module = self.module self.learner.dataset = self.history self.learning = True
def __init__(self, learner, **kwargs): LoggingAgent.__init__(self, learner.num_features, 1, **kwargs) self.learner = learner self.learner._behaviorPolicy = self._actionProbs self.reset()
def __init__(self, learner, **kwargs): LoggingAgent.__init__(self,2,1, **kwargs) self.learner = learner #self.reset() self.learning=True self.learner.dataset=self.history