Splendor-AI

Special Thanks

We would like to thank Prof. Nir Lipovetzky, an instructor in COMP90054 “AI Planning for Autonomy” course of the University of Melbourne. The course staff have organized a contest for autonomous agents for the game Splendor, developed by the students. We’ve contacted associate Prof. Nir Lipovetzky ( Github , Mail ) of Melbourne University and he provided us with their implementation of the game engine. From there we’ve started tweaking the game engine a bit for our needs (like adding the generatePredecessor method to SplendorGameRule ).

Contents:

• Installation of Splendor
  • Install Splendor using conda:
  • Install Splendor using pip:
  • Verify The Installation
• Run the game
  • Interactive mode
  • Specify Opponents
  • Without GUI (Textual Mode)
  • Using Our Agents
• Developing an Agent
  • Training Our Agents:
  • Training The Genetic Algorithm Agent:
  • Training The PPO Agent:
  • SplendorEnv - an OpenAI gym compatible simulator for the game Splendor
• Generate The Documentation
  • Automatic generation of documentation
  • Publishing the Documentation to Github Pages
  • Highly Autonomous Documentation Generation & Publication
• Development
  • Executation of Automatic Tools by Hand
  • Using Pre-Commit Git Hooks
• splendor
  • splendor package
• Agents
  • splendor.agents package