A set of Jupyter Notebooks which build up reinforcement learning techniques from scratch.
I've seen a lot of video series on RL, but none of them came with assignments that would ensure that I actually learned the material. This repository is designed as a set of RL 'assignments' to teach all of the fundamentals of reinforcement learning. It is a series of ipython notebooks with markdown instructions, and the implementation is left to the reader (my solutions are availiable too, but you really shouldn't look at them).
One thing to note is that at the start of this, I had only watched one lecture series:
https://www.youtube.com/watch?v=2pWv7GOvuf0&list=PLzuuYNsE1EZAXYR4FJ75jcJseBmo4KQ9-
Hopefully all that is required for you to learn RL yourself is to watch that series, and then complete this series of assignments, since I'll stop once I feel I have a fundamental understanding of RL.
My goal in learning generally is to be able to, by the end, design novel systems. This usually requires a fundamental understanding of how all of the components work, and then an intuitive grasp of how things fit together. The goal of this series of exercises is to gain that understanding.