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()