ray/rllib/tuned_examples/regression_tests/cartpole-dqn-param-noise-tf.yaml

cartpole-dqn-tf-w-param-noise:
    env: CartPole-v0
    run: DQN
    stop:
        episode_reward_mean: 150
        timesteps_total: 300000
    config:
        use_pytorch: false
        exploration_config:
            type: ParameterNoise
            random_timesteps: 10000
            initial_stddev: 1.0
        batch_mode: complete_episodes
        lr: 0.0008
        num_workers: 0
        model:
            fcnet_hiddens: [32, 32]
            fcnet_activation: tanh