https://github.com/virresh/pommerwarrior

Warrior Agent for Pommerman
https://github.com/virresh/pommerwarrior

Last synced: 10 months ago
JSON representation

Warrior Agent for Pommerman

• Host: GitHub
• URL: https://github.com/virresh/pommerwarrior
• Owner: virresh
• Created: 2018-11-22T04:26:54.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2019-02-16T15:43:06.000Z (over 7 years ago)
• Last Synced: 2025-03-12T15:15:49.656Z (over 1 year ago)
• Language: Python
• Size: 2.6 MB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
## Pommerwarrior

RL Agent developed as a course project for AI class. Current performance: the learnt disco agent beats the baseline SimpleAgent with a win ratio of 35%.

## Instructions to Run

1. Install OpenAI Gym from pip.

2. Follow [installation instructions for Pommerman](https://github.com/MultiAgentLearning/playground/tree/master/docs)

3. Run `src/Runner.py`

Step 3 launches 3 of our trained agents against the baseline agent. You may vary this composition by changing the runner script.

## Some Winning Plays

Red Agent (Top left) is our best performing Agent.  

Green Agent (Top right) is the baseline Agent.  

The other two are our suboptimal Agents.  

![demo-red-winning.gif](demo-red-winning.gif)