def main(args):
"""
train and save the DeepQ model, for the cartpole problem
:param args: (ArgumentParser) the input arguments
"""
env = gym.make("CartPole-v0")
model = DeepQ(
env=env,
policy=MlpPolicy,
learning_rate=1e-3,
buffer_size=50000,
exploration_fraction=0.1,
exploration_final_eps=0.02,
)
model.learn(total_timesteps=args.max_timesteps, callback=callback)
print("Saving model to cartpole_model.pkl")
model.save("cartpole_model.pkl")
def main(args):
"""
train and save the DeepQ model, for the mountain car problem
:param args: (ArgumentParser) the input arguments
"""
env = gym.make("MountainCar-v0")
# using layer norm policy here is important for parameter space noise!
model = DeepQ(policy=CustomPolicy,
env=env,
learning_rate=1e-3,
buffer_size=50000,
exploration_fraction=0.1,
exploration_final_eps=0.1,
param_noise=True)
model.learn(total_timesteps=args.max_timesteps)
print("Saving model to mountaincar_model.pkl")
model.save("mountaincar_model.pkl")