def test_write_upon_reset_true():
with helpers.tempdir() as temp:
env = gym.make('CartPole-v0')
# TODO: Fix Cartpole to not configure itself automatically
# assert not env._configured
env = Monitor(env, directory=temp, video_callable=False, write_upon_reset=True)
env.configure()
env.reset()
files = glob.glob(os.path.join(temp, '*'))
assert len(files) > 0, "Files: {}".format(files)
env.close()
files = glob.glob(os.path.join(temp, '*'))
assert len(files) > 0
def test_write_upon_reset_true():
with helpers.tempdir() as temp:
env = gym.make('CartPole-v0')
# TODO: Fix Cartpole to not configure itself automatically
# assert not env._configured
env = Monitor(env,
directory=temp,
video_callable=False,
write_upon_reset=True)
env.configure()
env.reset()
files = glob.glob(os.path.join(temp, '*'))
assert len(files) > 0, "Files: {}".format(files)
env.close()
files = glob.glob(os.path.join(temp, '*'))
assert len(files) > 0
gamma=0.8,
train_freq=1,
gradient_steps=1,
target_update_interval=50,
verbose=1,
tensorboard_log="highway_dqn/")
# Train the model
if TRAIN:
model.learn(total_timesteps=int(2e4))
model.save("highway_dqn/model")
del model
# Run the trained model and record video
model = DQN.load("highway_dqn/model", env=env)
env = Monitor(env, directory="highway_dqn/videos", video_callable=lambda e: True)
env.set_monitor(env)
env.configure({"simulation_frequency": 15}) # Higher FPS for rendering
for videos in range(10):
done = False
obs = env.reset()
while not done:
# Predict
action, _states = model.predict(obs, deterministic=True)
# Get reward
obs, reward, done, info = env.step(action)
# Render
env.render()
env.close()
