73b81f0b1909341c1f5633242c101498b12e84c9,chapter06/maximization_bias.py,,figure_6_7,#,103

Before Change

        for ep in range(0, episodes):
            left_counts_q[run, ep] = q_learning(q)
            left_counts_double_q[run, ep] = q_learning(q1, q2)
    left_counts_q = np.add.accumulate(left_counts_q, axis=1).mean(axis=0) / np.arange(1, episodes + 1)
    left_counts_double_q = np.add.accumulate(left_counts_double_q, axis=1).mean(axis=0) / np.arange(1, episodes + 1)

    plt.plot(left_counts_q, label="Q-Learning")
    plt.plot(left_counts_double_q, label="Double Q-Learning")

After Change

            left_counts_q[run, ep] = q_learning(q)
            left_counts_double_q[run, ep] = q_learning(q1, q2)
    left_counts_q = left_counts_q.mean(axis=0)
    left_counts_double_q = left_counts_double_q.mean(axis=0)

    plt.plot(left_counts_q, label="Q-Learning")
    plt.plot(left_counts_double_q, label="Double Q-Learning")
    plt.plot(np.ones(episodes) * 0.05, label="Optimal")

Instances