2048-api

A 2048 game api for training supervised learning (imitation learning) or reinforcement learning agents

Code structure

• game2048/: the main package.
  • 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.
• MyCode/: the main work to complete the task.
  • agent.py: the Agent I create as MyAgent.
  • mymodel.py: the model I build.
  • onehot.py: used to change to game board to a one_hot_encoded form.
  • train.py: the codes used to train the model.
  • trained_model.h5: the final trained model.
• 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

To define your own agents

from game2048.agents import Agent

class YourOwnAgent(Agent):

    def step(self):
        '''To define the agent's 1-step behavior given the `game`.
        You can find more instance in [`agents.py`](game2048/agents.py).
        
        :return direction: 0: left, 1: down, 2: right, 3: up
        '''
        direction = some_function(self.game)
        return direction

To compile the pre-defined ExpectiMax agent

cd game2048/expectimax
bash configure
make

To run the web app

python webapp.py

LICENSE

The code is under Apache-2.0 License.

For EE369 / EE228 students from SJTU

Please read course project requirements and description.