Deep Q-Networks in Python
In Conda environment, please replace python3
as python
.
For CartPole-v0:
python3 ./dqn.py --network_type DQN --env CartPole-v0 --gamma 0.99 --train_episode 400
For MountainCar-v0:
python3 ./dqn.py --network_type DQN --env MountainCar-v0 --gamma 1 --train_episode 1000
For CartPole-v0:
python3 ./dqn.py --network_type Dueling_DQN --env CartPole-v0 --gamma 0.99 --train_episode 500
For MountainCar-v0:
python3 ./dqn.py --network_type Dueling_DQN --env MountainCar-v0 --gamma 1 --train_episode 3000
For SpaceInvaders-v0:
python3 ./dqn.py --network_type Conv_DQN --env SpaceInvaders-v0 --gamma 0.99 --train_episode 100 --memory_size 1000000 --burn_in 50000 --target_update_frequency 10000