https://github.com/jihoonerd/continuous-control-with-deep-reinforcement-learning

📖 Paper: Continuous control with deep reinforcement learning 🕹️
https://github.com/jihoonerd/continuous-control-with-deep-reinforcement-learning

atari ddpg lunarlander-v2 pytorch reinforcement-learning

Last synced: about 2 months ago
JSON representation

📖 Paper: Continuous control with deep reinforcement learning 🕹️

• Host: GitHub
• URL: https://github.com/jihoonerd/continuous-control-with-deep-reinforcement-learning
• Owner: jihoonerd
• License: mit
• Created: 2021-02-01T03:27:23.000Z (almost 4 years ago)
• Default Branch: main
• Last Pushed: 2024-04-07T01:58:44.000Z (9 months ago)
• Last Synced: 2024-05-22T00:05:08.006Z (8 months ago)
• Topics: atari, ddpg, lunarlander-v2, pytorch, reinforcement-learning
• Language: Python
• Homepage:
• Size: 1.42 MB
• Stars: 1
• Watchers: 4
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Continuous Control with Deep Reinforcement Learning (DDPG)

This implements a reinforcement learning algorithm [DDPG](https://arxiv.org/abs/1509.02971). Using the methodology employed for DQN, DDPG can resolve continous action space environments.

Following shows that DDPG can solve one of the continuous action space environment in `OpenAI Gym`.

|Episode: 0|Episode: 500|Episode: 900|

|---|---|---|

|![eps-0](assets/eps-0.gif)|![eps-500](assets/eps-500.gif)|![eps-900](assets/eps-900.gif)|

## Score Graph for `LunarLanderContinuous-v2`



## Environments

* Pytorch 1.7

* Python 3.8

Please refer `requirements.txt` for python packages for this repo.