Reproducibility Challenge: Q-map: a Convolutional Approach for Goal-Oriented Reinforcement Learning

This repository pertains to the ICLR 2019 reproducibility challenge submission for the paper Q-map: a Convolutional Approach for Goal-Oriented Reinforcement Learning. [https://openreview.net/forum?id=rye7XnRqFm]

Getting Started

These instructions will get you a copy of the project up and running on your local machine.

Prerequisites

Pytorch (0.4.1)
Baselines (0.1.5)
Gym (0.10.5)
Gym Retro (0.6.0)

Installing

git clone https://github.com/fabiopardo/qmap.git
cd qmap  
pip install -e .  

Copy the SuperMarioAllStars-Snes folder to the retro/data/stable directory where Gym Retro is installed.
Finally clone this repository into the qmap folder \

Running

python train_mario.py  

For loading a previously saved model, simply pass the DQN or Q-Map files step value as load argument.
The training of the agent can also be accomplished by running the jupyter notebook train_mario.ipynb

Authors

• Shishir Sharma - (https://github.com/shishir13sharma)

Acknowledgments

• Code reproduced from https://github.com/fabiopardo/qmap