- Python
>= 3.6
- PIP
>= 20.2.4
pip install -r requirements.txt
- Make sure you have copied all models to
{project root}/models
directory (link) - Decide which agent you want to run. Available agents:
rand
- Random Agentdqn
- DQN Agent (Baselines3)a2c
- A2C Agent (Baselines3)ppo
- PPO Agent (Baselines3)dqn_custom
- DQN Agent (implemented with PyTorch)dqn_forgetting
- DQN Agent (implemented with PyTorch, shows agent performance after catastrophic forgetting)
- Decide how long the algorithm should sleep between time steps (default value is 0.05s)
- Decide how many episodes (full games) the algorithm should play (default value is 1)
- Run with
python run.py --agent <chosen agent> [--sleep <float>] [--episodes <int>]