Aananda-giri / game_of_29

game_of_29 a card game

Info

• An agent for game_of_29 a card game
• contains:
  • Engine: engine/Play.py game_of_29 playing engine
  • Agent: engine/python a sanic webserver
• developed for smartbot 2023 Challange

Todo:

- engine performance pickle & evaluate
- MCTS tree search
- deep Q learning

tutorials:

• pacman using mdp

• understanding mdp
• Essentials of Reinforcement Learning

resources

Industry graded RL libraries:

• baseline openai (newer)

• stable baselines (older)

• rllib : ray library

• introduction-to-deep-q-networks

• deep-double-q

• dqn vs double dqn

to run:

python3 29_engine.py

[1st MILESTONE: study phase complete] ( High Level Design and Low Level Design)

todo

• ISMCTS
• Maximal Bipartite Matching for distributing cards
• implement in c++ for speed
• store table states
• 1v3 game
• 1v1 game