carlolucibello / deeprlexamples.jl

Deep Reinforcement Learning examples using Julia

DeepRLexamples

This repo provides examples of deep reinforcement learning in julia (v1.0 and above) using Knet deep learning library and OpenAI Gym. Contributions are very welcome!

Installation

Install the gym environment for python

pip install --user gym[atari]

and the julia packages Gym.jl and Knet

] add Gym Knet

You are now ready to run any of the examples in the repo. You can clone the whole repo with

git clone https://github.com/CarloLucibello/DeepRLexamples.jl

Usage

include("actor_critic_pong.jl")
main(seed=17, episodes=1000, lr=1e-2, render=true, infotime=50)

Examples

• reinforce_cartpole.jl: reinforce algorithm with a multi-layer perceptron. CPU only.

• actor_critic_cartpole.jl: actor critic algorithm with a multi-layer perceptron. CPU only.

• actor_critic_pong.jl: actor critic algorithm with a convolutional neural network. Following Karphaty's blog entry, but using actor-critic instead of simple police gradient. Also, a convolutional neural network instead of a multi-layer perceptron. Runs on both CPU and GPU.

deeprlexamples.jl's Issues

missing actors

• A3C
• A2C
  https://blog.openai.com/baselines-acktr-a2c/
• ACKTR
  https://blog.openai.com/baselines-acktr-a2c/
• DQN
• Double DQN
  https://arxiv.org/abs/1509.06461

saying hi

Hi @etotheipluspi @Evizero @denizyuret @ilkerkesen @ozanarkancan @sglyon,
I've been writing some deep reinforcement examples using julia and knet and I'm pinging you just to let you know, since i think you might be interested and also since I used some of your code (especially @ilkerkesen's).

Reinforcement learning situation in julia is kind of fragmented, and what could have been the reference package, https://github.com/JuliaML/Reinforce.jl, seems unfinished and unmaintened.

I though It would be nice to have a julia repo with neat examples and as little as possible external dependencies. This repo could be moved to the JuliaML org at a certain point.

Bye,
Carlo

some related julia repos:
https://github.com/ilkerkesen/rl
https://github.com/JuliaML/Reinforce.jl
https://github.com/ozanarkancan/ReinforcementLearning.jl
https://github.com/sisl/DeepRL.jl
https://github.com/JuliaPOMDP/POMDPs.jl
https://github.com/Evizero/ReinforcementLearning.jl
https://github.com/JuliaPOMDP/POMDPReinforce.jl

and discussions:
JuliaML/Reinforce.jl#10
JuliaML/Reinforce.jl#5
JuliaML/Reinforce.jl#2
JuliaPOMDP/POMDPs.jl#142
https://github.com/Evizero/ReinforcementLearning.jl/blob/master/docs/src/devdocs/design.md