I am new to Julia. I want to explore reinforcement learning using julia. But there are so many packages available for reinforcement learning using julia.
Can somebody guide which package should I use for example to test atari - breakout game.
Also, is there any tutorial available to understand the steps/flow/other required packages involved in building any application using julia’s reinforcement learning package.
As I searched alot but not able to understand how to start building and understanding RL game uisng Julia.
I am no reinforcement learning expert, but it seems that the AtariAlgos.jl package provides Atari games as Reinforce.jl compatible environments.
For example see here https://github.com/JuliaML/AtariAlgos.jl#example .
Here is an example of using DQN from ReinforcementLearning.jl to learn to play an atari game.
However, we (actually mostly @findmyway ) are currently working on a major refactoring, with ReinforcementLearningEnvironments.jl the main package for environments and ReinforcementLearning.jl the one for algorithms. You may also find the ReinforcementLearningAnIntroduction.jl package interesting as it shows how to use these packages. But the current, not entirely stable and not yet well documented master branches work only on julia 1.3. We still need a bit of time to make everything smooth, but you are more than welcome to try out things!
Thanks to all of you for replying.
I have gone through the following:
https://github.com/JuliaML/AtariAlgos.jl. But is it training and making it learn how to play or just randomly running few episodes?
But this code is not working. I have installed Julia version 1.0.5 but this code is throwing errors. Attached the error images.
I want to train an agent using Reinforcement learning in Julia to play Atari breakout. Would appreciate if someone can suggest the steps and which pakcage to refer to.
I am not able to understand which package to use to start with as there are so many packages available for RL using Julia and each one provides different things.
I am having a hard time in relating these packages and the order in which to use them and where to start from.