A SUPER QUICK way to train Mountain Car using DQN

Reinforcement Learning DQN - using OpenAI gym Mountain Car

• Keras
• gym

The training will be done in at most 6 minutes! (After about 300 episodes the network will converge

The program in the video is running in macOS(Macbook Air) , and it only took 4.1 minutes to finish training. No GPU used.

Using GPU

You can use codes:

import tensorflow as tf
import keras
config = tf.ConfigProto( device_count = {'GPU': 2 , 'CPU': 1} ) 
sess = tf.Session(config=config) 
keras.backend.set_session(sess)

and change the trainFromBuffer function to Boost

#self.trainFromBuffer()
self.trainFromBuffer_Boost()

I used a workstation to run GPU version. It took about 2mins to finish training.

How to test

run testMountainCar.py

change the file path in

#load the network
model=models.load_model('your model filepath')

Then you can see how the car plays the game.