pendulum-td3-tf:
    env: Pendulum-v0
    run: TD3
    stop:
        use_pytorch: false
        episode_reward_mean: -900
        timesteps_total: 100000