Ray tune user guide for hyperparameter optimization

This tutorial details the use of the Ray tune for hyperparameter optimization. There have been a lot of changes in the Ray ecosystem after ray 2.2.0 was released. So the tutorial covers all the changes in the latest version and all the functionalities have been encapsulated to abstract away the complexities. The tutorial is structured as follows: (i)It starts with understanding the code, (ii)Understanding the meaning behind each attribute, and then finally running it.

This is the migration doc for Ray from 1.* to 2.* is here

In my previous article, I went through a tutorial guide on using Ray tune and RLlib for hyperparameter tuning of a Stock trading environment using Reinforcement learning. Now, ray built by ANYSCALE, has released Ray 2.2.0, and they have simplified the different implementation details. So let’s start with the implementation details.

PARAMETERS REQUIRED

• TRAINABLE: It is actually any trainer which will accept a config parameter. The ray documentation describes the trainable class as follows

Taken from here

• You can see that the session report is used to report the different metrics which…