Skip to content

Files

Latest commit

5382a21 · Feb 10, 2020

History

History

pong

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
Feb 10, 2020
Feb 10, 2020
Feb 10, 2020

README.md

Pong

Trains a Pong agent using policy gradients on OpenAI's gym. This code was copied from Andrej Karpathy's Deep Reinforcement Learning: Pong from Pixels, and almost all changes to the code were for cosmetic purposes. Please refere to Karpathy's walkthrough to learn more about the implementation!

Usage

python3 pong.py

Set resume = True in pong.py if you want to continue training the agent where it was left off in model.p, otherwise set resume = False to start the agent training from scratch.

Output

Resuming model 'model.p'...
ep 0: game finished, reward: 1.000000
ep 0: game finished, reward: 1.000000
ep 0: game finished, reward: 1.000000
ep 0: game finished, reward: -1.000000
ep 0: game finished, reward: 1.000000
ep 0: game finished, reward: -1.000000
...
...
...