continuous-control

This repo is an implementation of the research paper "Continuous control with deep reinforcement learning"(DDPG).

Demo

train.py: train the agent
random_process.py: the random noise file(Ornstein-Uhlenbeck process here)
ddpg.py: the agent(the actor and critic networks)
display.py: load the trained model and see how the agent interact with the environment (Here are the pretrained models on HalfCheetah-v4, LunarLanderContinuous-v2 and Pendulum-v1 from this code https://drive.google.com/drive/u/0/folders/1gGhH0ad7OT9_tuMAaJq36lVR6S-C1Ag6)

Name		Name	Last commit message	Last commit date
Latest commit History 14 Commits
.gitignore		.gitignore
README.md		README.md
ddpg.py		ddpg.py
display.py		display.py
random_process.py		random_process.py
replay_buffer.py		replay_buffer.py
requirements.txt		requirements.txt
train.py		train.py