This Python Package can be used to train an agent to play various Atari Games (OpenAI Gym) using Deep Q-Learning.
$ pip install -r requirements.txt
$ python -m dqn --agent <agent_name> --mode <train/simulate>
--agent
: Specify the agent to be used (Refer to Agents).--mode
: Train an agent (train
) or simulate a trained agent (simulate
).
More information here.
The following simulation is 3 episodes of a trained CartPole agent receiving a mean reward of 200.0
after training for 1000 episodes.
$ python -m dqn --agent cartpole --mode train
$ python -m dqn --agent cartpole --mode simulate
In-progress