项目作者: ShijieVVu

项目描述 :
高级语言: Python
项目地址: git://github.com/ShijieVVu/Model-Free-RL-Framework.git
创建时间: 2018-03-08T14:39:23Z
项目社区:https://github.com/ShijieVVu/Model-Free-RL-Framework

开源协议:

下载


Model-free Reinforcement Learning Models with Experience Replay

======

Installation

Install dependencies
pip install gym
pip install h5py
pip install tensorflow

Description

Implementation of linear, MLP and Duel DQN model with experience replay to solve CartPole-v0 and MountainCar-v0 environment(MLP and Duel DQN can solve both challenges).
SpaceInvaders could achieve 300 or so scores.

Run

To solve the CartPole-v0 environment using Duel DQN with experience replay, use
python CartPole_q4.py
To solve the MountainCar-v0 environment using MLP with experience replay, use
python MountainCar_q3.py