https://github.com/saharan/webdl-ddpg

Deep Deterministic Policy Gradient with WebGL and CUDA
https://github.com/saharan/webdl-ddpg

deep-learning deep-reinforcement-learning

Last synced: 11 days ago
JSON representation

Deep Deterministic Policy Gradient with WebGL and CUDA

• Host: GitHub
• URL: https://github.com/saharan/webdl-ddpg
• Owner: saharan
• License: mit
• Created: 2018-06-30T05:51:00.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2018-06-30T10:21:38.000Z (over 6 years ago)
• Last Synced: 2024-11-06T08:13:01.667Z (2 months ago)
• Topics: deep-learning, deep-reinforcement-learning
• Language: Haxe
• Size: 15.8 MB
• Stars: 20
• Watchers: 5
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
WebDL-DDPG

---

A library for deep reinforcement learning. The article about this library [is here](http://el-ement.com/blog/2018/06/30/full-ddpg/) (Japanese).

## Features

* Supports GPU (WebGL for JavaScript platform and CUDA for Python platform)

* DDPG (Deep Deterministic Policy Gradient) is implemented

## Main classes

* [src/RobotDDPGTest.hx](./src/RobotDDPGTest.hx) for playing a learned model of bipedal walking robot (JS)

* [src/RobotDDPGCudaTest.hx](./src/RobotDDPGCudaTest.hx) for learning and exporting models of bipedal walking robot (Python)

* [src/PendulumTest.hx](./src/PendulumTest.hx) for learning and playing an inverted pendulum swingup problem (JS)

## License

The MIT License