- Dependencies are in
requirements.txt
- Best performances (episodes recordings) are in
episodes/
- Pretrained model weights are in
pretrained_models/
- Create new virtual environment
python3 -m venv ./ venv
- Install all dependencies
pip3 install -r requirements.txt
- Activate the newly created env
source venv/bin/activate
- Test demo (with saved pretrained model weights and does not trigger episode recordings)
python3 demo.py
-
(Extra) Choose which agent you want to test (Notice: these commands will trigger the recording of episodes and will save the newly trained model weights)
- Random Agent (episode recordings are in
episodes/random_agent
)python3 random_agent.py
- Deep Q-Network CartPole agent (episode recordings are in
episodes/dqn_agent
)python3 dqn_agent.py
- Deep Q-Convolutional Network CartPole agent (episode recordings are in
episodes/conv_dqn_agent
)python3 conv_dqn_agent.py
- Deep Q-Convolutional Network MineRL agent
python3 minerl_agent.py
- Random Agent (episode recordings are in
Finally, don't forget to deactivate the virtual environment after testing
deactivate
- Khaled ABDRABO p1713323
- Jean BRIGNONE p1709655