config['gamma'], config['max_steps'],
config['green_duration'],
config['yellow_duration'], config['num_states'],
config['num_actions'], config['training_epochs'])
episode = 0
timestamp_start = datetime.datetime.now()
gui_episodes = config['gui_episodes']
while episode < config['total_episodes']:
print('\n----- Episode', str(episode + 1), 'of',
str(config['total_episodes']))
epsilon = 1.0 - (
episode / config['total_episodes']
) # set the epsilon for this episode according to epsilon-greedy policy
simulation_time, training_time = Simulation.run(
episode, epsilon, episode in gui_episodes) # run the simulation
print('Simulation time:', simulation_time, 's - Training time:',
training_time, 's - Total:',
round(simulation_time + training_time, 1), 's')
episode += 1
print("\n----- Start time:", timestamp_start)
print("----- End time:", datetime.datetime.now())
print("----- Session info saved at:", path)
Model.save_model(path)
copyfile(src='training_settings.ini',
dst=os.path.join(path, 'training_settings.ini'))
Visualization.save_data_and_plot(data=Simulation.reward_store,
config['gamma'], config['max_steps'],
config['green_duration'],
config['yellow_duration'], config['num_states'],
config['num_actions'], config['training_epochs'])
episode = 0
timestamp_start = datetime.datetime.now()
# Start of simulation
while episode < config['total_episodes']:
print('\n----- Episode', str(episode + 1), 'of',
str(config['total_episodes']))
epsilon = 1.0 - (
episode / config['total_episodes']
) # set the epsilon for this episode according to epsilon-greedy policy
simulation_time, training_time = Simulation.run(
episode, epsilon) # run the simulation
print('Simulation time:', simulation_time, 's - Training time:',
training_time, 's - Total:',
round(simulation_time + training_time, 1), 's')
episode += 1
print("\n----- Start time:", timestamp_start)
print("----- End time:", datetime.datetime.now())
print("----- Session info saved at:", path)
Model.save_model(path)
copyfile(src='training_settings.ini',
dst=os.path.join(path, 'training_settings.ini'))
Visualization.save_data_and_plot(data=Simulation.reward_store,