In this tutorial I will show you how RUDDER can be applied step by step and how a reward redistribution model can be implemented using PyTorch.
You may use it as a quick-guide to apply RUDDER in your RL setting and to pre-assess if RUDDER might lead to improvements for your task beforehand. This code should be runnable on common CPUs in reasonable time.
Links to further RUDDER code, our blog, and paper can be found at our RUDDER repo.