Skip to content

The pytorch implementation of DGN on grid world and Starcraft

License

Notifications You must be signed in to change notification settings

jiechuanjiang/pytorch_DGN

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

65 Commits
 
 
 
 
 
 
 
 

Repository files navigation

pytorch_DGN

This is a pytorch implementation of DGN.

Surviving

Here, we propose a multi-agent environment Surviving to evaluate DGN in large-scale situation. There are 100 agents interacting with the environment. Each agent corresponds to one grid and has a vary limited local observation that contains a square view with 3 x 3 grids centered at the agent. The agent could communicate with the neighboring agents in the square region with 7 x 7 grids. At each timestep, each agent can move to one of four neighboring grids or eat the food at its location. Agents begin with 10 health and lose 1 health per step. Eating food could increase the health. If the agent hits 0 health, it gets a reward -0.2, otherwise the reward is 0.4.

Surviving

Citation

If you are using the codes, please cite our paper.

Jiechuan Jiang, Chen Dun, Tiejun Huang, and Zongqing Lu. Graph convolutional reinforcement learning. ICLR'20.

@inproceedings{jiang2020graph,
    	title={Graph Convolutional Reinforcement Learning},
    	author={Jiang, Jiechuan and Dun, Chen and Huang, Tiejun and Lu, Zongqing},
    	booktitle={ICLR},
    	year={2020}
}

About

The pytorch implementation of DGN on grid world and Starcraft

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages