10337c8271537f0761486f9bf051ccffb5cb1838,examples/td.py,,experiment,#,9

Before Change

    //from keras.callbacks import EarlyStopping
    //config["algorithm"]["params"]["fit_params"]["callbacks"] =\
    //    [EarlyStopping("loss", patience=20, min_delta=1e-4)]
    alg = get_algorithm(config["algorithm"]["name"],
                        agent,
                        mdp,
                        **config["algorithm"]["params"])

After Change

    mdp = CarOnHill()

    // Policy
    epsilon = Parameter(value=1)
    discrete_actions = mdp.action_space.values
    pi = EpsGreedy(epsilon=epsilon, discrete_actions=discrete_actions)

    // Approximator

Instances