Reinforcement Learning with pytorch

This repository implements the classic and state-of-the-art deep reinforcement learning algorithms using pytorch. The goal of this repository is to provide an easy to read pytorch reinforcement learning implementation.

What is included?

Currently this respository includes the agents:

• Deep Q-learning [1]
• Deep Deterministic Policy Gradient (DDPG) [2]

Requirements

• Python 3.7
• gym >= 0.10
• pytorch >= 0.4

Installation

pip install -r requirements.txt 

In case failing with installation:

• Install gym environment

pip install gym 

• Install pytorch

please go to official webisite to install it: https://pytorch.org/

Recommend use Anaconda Virtual Environment to manage your packages

Deep Q-learning

The result of the DQN for training the cart pole (CartPole-v0) is upleaded here

First clone the repository

git clone https://github.com/shayantaherian/Reinforcement-Learning.git

Then move to the directory

cd Deep Q-learning

To start training, run the following command

python main.py

Deep Deterministic Policy Gradient (DDPG)

The result of the DQN for training the pendulum (Pendulum-v0) is upleaded here

First clone the repository

git clone https://github.com/shayantaherian/Reinforcement-Learning.git

Then move to the directory

cd DDPG

To start training, run the following command

python main.py

To test the result for generalization purposes run

python Test_result.py

References

1. DeepQLearning
2. Deeplizard
3. Deep Deterministic Policy Gradients