cartpole-appo-vtrace:
    env: CartPole-v0
    run: APPO
    stop:
        episode_reward_mean: 180
        timesteps_total: 200000
    config:
        # Works for both torch and tf.
        framework: tf
        num_envs_per_worker: 5
        num_workers: 4
        num_gpus: 0
        observation_filter: MeanStdFilter
        num_sgd_iter: 1
        vf_loss_coeff: 0.01
        vtrace: true
        vtrace_drop_last_ts: false
        model:
            fcnet_hiddens: [32]
            fcnet_activation: linear
            vf_share_layers: true