Complete the following exercises on your own:
- Alter the GridWorldBrain parameters in the trainer_config.yaml file and run further training sessions to explore the effect of changing the parameters.
- Build the 3DBalls environment and train it with the learn.py PPO algorithm using an external brain.
- Alter the parameters of the trainer_config.yaml file for the Ball3DBrain, run the simulation again, and view the results with TensorBoard.
Be sure to take some time to run through a few examples and configure some of the training parameters. Understanding what effect these hyperparameters can have on the impact of model training can be critical for you to successfully train working models.