Skip to content

ridersofrohan/maRLio

 
 

Repository files navigation

Reinforcement Learning for Super Mario Level 1


Installation ============

Uses (https://github.com/koltafrickenfer/gym-super-mario)

brew install fceux

virtualenv env
source env/bin/activate

pip install -r requrements.txt

cp -r mario_env/envs/kolta_gym_super_mario env/lib/python3.6/site-packages/gym/envs/
cp -r mario_env/envs/__init__.py env/lib/python3.6/site-packages/gym/envs/__init__.py
cp -r mario_env/scoreboard/__init__.py env/lib/python3.6/site-packages/gym/scoreboard/__init__.py

To load and run the environments, run

import gym
env = gym.make('SuperMarioBros-1-1-v0')
observation = env.reset()
for _ in range(1000):
    env.render()
    action = env.action_space.sample() # your agent here (this takes random actions)
    observation, reward, done, info = env.step(action)

About

Reinforcement Learning applied to the first level of Super Mario Bros..

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Jupyter Notebook 85.9%
  • Python 11.5%
  • Lua 2.6%