martin-xia0/2048_Planning

2048 Agent using Monte Carlo method, can reach 4096

2048_Planning

2048 Agent using Monte Carlo method, can reach 4096. It has a cnn layer to extract checkerboard feature and a lstm layer to make logic decision, and utilize a strong model to train itself.

Code structure

• game2048/: the main package.
  • montecarlo.py: the main model for this planning method
  • game.py: the core 2048 Game class.
  • agents.py: the Agent class with instances.
  • displays.py: the Display class with instances, to show the Game state.
  • expectimax/: a powerful ExpectiMax agent by here.
• explore.ipynb: introduce how to use the Agent, Display and Game.
• static/: frontend assets (based on Vue.js) for web app.
• webapp.py: run the web app (backend) demo.
• evaluate.py: evaluate your self-defined agent.

Requirements

• code only tested on linux system (ubuntu 16.04)
• Python 3 (Anaconda 3.6.3 specifically) with numpy and flask

Run the web app

python3 webapp.py

Then it automatically run the planing agent. After a few minutes, you will get the result.

Result