# Pendulum SAC can attain -150+ reward in 6-7k
# Configurations are the similar to original softlearning/sac codebase
pendulum-sac:
    env: Pendulum-v1
    run: SAC
    stop:
        episode_reward_mean: -250
        timesteps_total: 10000
    config:
        # Works for both torch and tf.
        framework: tf
        horizon: 200
        soft_horizon: false
        q_model_config:
          fcnet_activation: relu
          fcnet_hiddens: [256, 256]
        policy_model_config:
          fcnet_activation: relu
          fcnet_hiddens: [256, 256]
        tau: 0.005
        target_entropy: auto
        no_done_at_end: true
        n_step: 1
        rollout_fragment_length: 1
        train_batch_size: 256
        target_network_update_freq: 1
        min_sample_timesteps_per_reporting: 1000
        replay_buffer_config:
          type: MultiAgentPrioritizedReplayBuffer
          learning_starts: 256
        optimization:
          actor_learning_rate: 0.0003
          critic_learning_rate: 0.0003
          entropy_learning_rate: 0.0003
        num_workers: 0
        num_gpus: 0
        metrics_smoothing_episodes: 5