The idea of this project is to develop a simple actor-critic reinforcemente learning model (A3C) in tensorflow that works in envoironments with continous actions.
Descritpion of the files:
- cart_pole: notebook that contains my implementation in tensorflow of the actor_critic model for discrete actions (both using single and multiple envoiroment), tested in the cart_pole envoiroment
- pendulum: notebook that contains my implementation of the actor_critic model for continous actions, tested in the pendulum envoiroment
- multiprocessing_env.py: contains the code from openai gym to run several envoiroments (several agents) in parallel