-
Download this package from
git clone [email protected]:lightDRL/lightDRL.git
or download this zip -
Install python requirements by following command
pip install -r requirements.txt
# if fail, try to use sudo
sudo pip install -r requirements.txt
Test example which you want to play, for example, play gym
python examples/gym_basic/gym_basic.py examples/gym_basic/DDPG_CartPole-v0.yaml
Or you can
cd examples/gym_basic
python gym_basic.py DDPG_CartPole-v0.yaml
python server.py
python examples/gym_basic/gym_basic_conn.py examples/gym_basic/DDPG_CartPole-v0.yaml
Or you can
cd examples/gym_basic
python ../../server.py
python gym_basic_conn.py DDPG.yaml
Support NNcomponent, you can modify network by config.yaml
Update Q-learning to more general Play maze with Q-learning Add cfg['misc']['ep_max_step'], if done = True and self.ep_use_step >= self.ep_max_step -> train_done = False
python examples/maze/maze.py examples/maze/Q-learning.yaml
Support DQN! Now support DQN, DDPG, and Q-learning
Play maze with DQN
python examples/maze/maze.py examples/maze/DQN.yaml
Support Q-learning!
python examples/gym_basic/gym_basic.py examples/gym_basic/Q-learning_Taxi-v2.yaml
Change framework
Now, support standalone & server-client version !!