HMT-GAME-1-AGENTS

This repository implements a low fidelity version of HMT Game 1 (Dice Adventure), a custom OpenAI Gym environment for Dice Adventure, and training code for agents to connect and learn to play the game. This uses Stable Baselines. This project is currently a work in progress. I worked on creating agents for this project with different playstyles.

Install required packages

  pip install -r requirements.txt

Run the PPO model to play Dice Adventure

  python training.py

In another terminal, start up the TensorBoard server to view training results

  tensorboard --logdir monitoring/dice_adventure_tensorboard/