https://github.com/takuseno/dqn-sokushukai

Sample DQN code for 速習会 in Wantedly
https://github.com/takuseno/dqn-sokushukai

dqn reinforcement-learning

Last synced: about 1 month ago
JSON representation

Sample DQN code for 速習会 in Wantedly

• Host: GitHub
• URL: https://github.com/takuseno/dqn-sokushukai
• Owner: takuseno
• Created: 2017-06-17T08:19:26.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2017-07-01T15:22:25.000Z (over 7 years ago)
• Last Synced: 2024-10-23T04:06:32.525Z (2 months ago)
• Topics: dqn, reinforcement-learning
• Language: Python
• Size: 12.7 KB
• Stars: 1
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
## dqn-sokushukai

Sample DQN code for 速習会 in Wantedly.

`dqn.py` is based on original [dqn.py](https://github.com/chainer/chainerrl/blob/master/chainerrl/agents/dqn.py) in chainerrl without features not talken in the lecture.

## requirements

- Python3

## dependencies

- chainer==2.0.0

- gym[atari]

- chainerrl==0.2.0

- OpenCV3

## ussage

### training

```

$ python train.py --gpu {0 or -1} --render --final-steps 10000000

```

### playing

```

$ python play.py --gpu {0 or -1} --render --load {path of models}

```