https://github.com/hamidzr/q-learning
A Deep Q Learning framework to train an agent to play simple board games w/o the help of CNNs
https://github.com/hamidzr/q-learning
Last synced: 3 months ago
JSON representation
A Deep Q Learning framework to train an agent to play simple board games w/o the help of CNNs
- Host: GitHub
- URL: https://github.com/hamidzr/q-learning
- Owner: hamidzr
- Created: 2018-04-13T15:05:00.000Z (about 7 years ago)
- Default Branch: master
- Last Pushed: 2018-05-01T04:44:54.000Z (about 7 years ago)
- Last Synced: 2025-01-21T00:45:44.201Z (4 months ago)
- Language: Python
- Homepage:
- Size: 225 KB
- Stars: 2
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Deep Q Learning
## TODO
- add pong and snake games
- pull the step and state logc out of the game(s)
- support for AI vs AI
- optimize the model, other archtectures (ongoing)
- port the model to browser. (tfjs?)
- browser to snap
- save the weights
- load the weights
- easy definitoin of actions, state and rewards
- reward clipping
- more games: frogger, catching games, pong, openai games, grid world
- checkout games from [qlearning4k](https://github.com/farizrahman4u/qlearning4k)
## Resources
- Mnih, Volodymyr, et al. "Playing atari with deep reinforcement learning." arXiv preprint arXiv:1312.5602 (2013)
- Different game environemtns form OpenAi's [Gym](https://gym.openai.com/)
- Base DQN [link](https://keon.io/deep-q-learning/)
- Base TicTacToe game [link](https://TODO)
- Base Snake game [gist](https://gist.githubusercontent.com/HonzaKral/833ee2b30231c53ec78e/raw/c4492fe1bf594c1c5d7194e9a4016213f15e327b/snake.py)
- Base Connectn game [gist](https://gist.github.com/poke/6934842)
- Base Pong game [github repo](https://github.com/kidscancode/intro-python-code)