Skip to content

Simple Example A3C Reinforcement Learning Algorithm in Tensorflow

Notifications You must be signed in to change notification settings

kkweon/A3C-Tensorflow

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Asynchronous Methods for Deep Reinforcement Learning (A3C)


After training for 6 hours
  • A3C implementation in TensorFlow
  • There are two versions
    • A3C using threads
    • A3C using distributed tensorflow
  • Tested on "Pong-v0" which is a stochastic environment due to random frame skipping
  • Original paper: https://arxiv.org/abs/1602.01783

Tensorboard

Episode length

  • The time length of an episode increases as it starts to play neck and neck against the AI
  • After an hour, the trained agent begins to beat the AI easily

Episode reward

  • The maximum score possible is 21

About

Simple Example A3C Reinforcement Learning Algorithm in Tensorflow

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published