# This configuration can expect to reach -160 reward in 10k-20k timesteps.
pendulum-ddpg:
    env: Pendulum-v0
    run: DDPG
    stop:
        episode_reward_mean: -500
        timesteps_total: 100000
    config:
        # Works for both torch and tf.
        framework: tf
        # === Model ===
        actor_hiddens: [64, 64]
        critic_hiddens: [64, 64]
        n_step: 1
        model: {}
        gamma: 0.99

        # === Exploration ===
        exploration_config:
            type: "OrnsteinUhlenbeckNoise"
            scale_timesteps: 10000
            initial_scale: 1.0
            final_scale: 0.02
            ou_base_scale: 0.1
            ou_theta: 0.15
            ou_sigma: 0.2

        timesteps_per_iteration: 600
        target_network_update_freq: 0
        tau: 0.001

        # === Replay buffer ===
        buffer_size: 10000
        prioritized_replay: True
        prioritized_replay_alpha: 0.6
        prioritized_replay_beta: 0.4
        prioritized_replay_eps: 0.000001
        clip_rewards: False

        # === Optimization ===
        actor_lr: 0.001
        critic_lr: 0.001
        use_huber: True
        huber_threshold: 1.0
        l2_reg: 0.000001
        learning_starts: 500
        rollout_fragment_length: 1
        train_batch_size: 64

        # === Parallelism ===
        num_workers: 0
        worker_side_prioritization: False