Reinforcement Learning Library

This repository aims to contain the latest reinforcement learning algorithms implemented using Tensorflow, Keras and OpenAI Gym.

Currently, A3C has been implemented.

Requirements

• Python 3.5

pip install -r requirements.txt

Usage

agent = A3CAgent(num_actions, lambda: model)
agent.train(env_name)

Tensorboard Logging

tensorboard --logdir=out --reload_interval=2

Sources:

• https://medium.com/emergent-future/simple-reinforcement-learning-with-tensorflow-part-8-asynchronous-actor-critic-agents-a3c-c88f72a5e9f2