Examples

This page lists detailled examples demonstrating how to use this package. They are meant to cover basic use cases.

• Cross evaluating random players
  • Getting something to run
  • Creating random players
  • Cross evaluating players
• Creating a simple max damage player
  • Creating a player
  • Creating a choose_move method
  • Method signature
  • Selecting a move
  • Returning a choice
  • Running and testing our agent
• Reinforcement learning with the OpenAI Gym wrapper
  • Implementing rewards and observations
    • Defining observations
    • Defining rewards
    • Defining our custom class
    • Instantiating and testing a player
    • Instantiating train environment and evaluation environment
  • Creating a DQN with keras-rl
    • Defining a base model
    • Defining the DQN
  • Training the model
  • Evaluating the model
    • Simple win rate evaluation
    • Use provided evaluate_player method
    • Use provided cross_evaluate method
  • Final result
• Adapting the max player to gen 8 OU and managing team preview
  • Team Preview management
  • Specifying a team
    • Packed format
    • Showdown format
    • Attributing a team to an agent
  • Running and testing our agent
• Creating a custom teambuilder
  • Creating a custom Teambuilder
    • Class definition
    • Instanciation
    • Launching battles
• Connecting to showdown and challenging humans
  • Connecting your agent to showdown
  • Challenging a human player
  • Accepting challenges from human players
  • Playing on the ladder