Exercises

As we progress through this book, the exercises at the end of each chapter will be more directed toward providing you with agent training experience. Training RL agents not only requires a fair amount of patience but also intuition on how to spot whether something is wrong or right. That only comes with training experience, so use the following exercises to learn that:

1. Open example Chapter_4_2.py and change the gridSize variable to see what effect this has on convergence.
2. Open example Chapter_4_2.py and tune the hyperparameters for alpha and gamma. Try to find the optimum values for both. This will require you to run the example multiple times.
3. Open example Chapter_4_2.py and change the number of episodes, up or down. See what effect a large number of episodes, such as 100,000 or 1,000,000, has on training.
4. Tune the learning_rate and gamma hyperparameters in example...