Training and testing the DDPG on Pendulum-v0

We will now train the preceding DDPG code on Pendulum-v0. To train the DDPG agent, simply type the following in the command line at the same level as the rest of the code:

python ddpg.py

This will start the training:

 
{'actor_lr': 0.0001,
 'buffer_size': 1000000,
 'critic_lr': 0.001,
 'env': 'Pendulum-v0',
 'gamma': 0.99,
 'max_episode_len': 1000,
 'max_episodes': 250,
 'minibatch_size': 64,
 'mode': 'train',
 'random_seed': 258,
 'render_env': False,
 'tau': 0.001}
.
.
.
2019-03-03 17:23:10.529725: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:300] kernel version seems to match DSO: 384.130.0
| Episode: 0 | Reward: -7981 | Qmax: -6.4859
| Episode: 1 | Reward: -7466 | Qmax: -10.1758
| Episode: 2 | Reward...