示例#1
0
文件: td.py 项目: nixworks/mushroom
    def __init__(self, policy, mdp_info, params):
        self.Q = Table(mdp_info.size)
        self._lambda = params['algorithm_params']['lambda']

        trace = params.get('algorithm_params', 'replacing')
        self.e = EligibilityTrace(self.Q.shape, trace)
        super(SARSALambdaDiscrete, self).__init__(self.Q, policy, mdp_info,
                                                  params)
示例#2
0
    def __init__(self,
                 policy,
                 mdp_info,
                 learning_rate,
                 lambda_coeff,
                 trace='replacing'):
        """
        Constructor.

        Args:
            lambda_coeff (float): eligibility trace coefficient;
            trace (str, 'replacing'): type of eligibility trace to use.

        """
        self.Q = Table(mdp_info.size)
        self._lambda = lambda_coeff

        self.e = EligibilityTrace(self.Q.shape, trace)
        super().__init__(self.Q, policy, mdp_info, learning_rate)