def test_loop(phenotypes, iteration):
"""Test the phenotypes in a completely different environment."""
# Create a validation environment instance
test = TestModel(100, width, height, 10, iter=iteration)
# Build the environment
test.build_environment_from_json()
# Create the agents in the environment from the sampled behaviors
test.create_agents(phenotypes=phenotypes)
# Store the initial result
testresults = SimulationResults(test.pname, test.connect,
test.sn, test.stepcnt,
test.foraging_percent(), phenotypes[0])
# Save the phenotype to a json file
testresults.save_phenotype()
# Save the data in a result csv file
testresults.save_to_file()
# Save the phenotype of json file
phenotype_to_json(test.pname, test.runid, phenotypes)
# Execute the BT in the environment
for i in range(iteration):
test.step()
testresults = SimulationResults(test.pname, test.connect,
test.sn, test.stepcnt,
test.foraging_percent(), phenotypes[0])
testresults.save_to_file()
# Plot the result in the graph
graph = GraphACC(test.pname, 'simulation.csv')
graph.gen_plot()
xlabel='Episode',
ylabel='Cumulative delay (s)')
viz.save_data_and_plot(data=sim._avg_queue_length_store,
filename='queue',
xlabel='Episode',
ylabel='Average queue length (vehicles)')
#%% Testing Simulation
config = import_train_configuration(config_file="training_settings.ini")
sumo_cmd = set_sumo(False, config['sumocfg_file_name'],
config['max_steps'])
TrafficGen = TrafficGenerator(config['max_steps'],
config['n_cars_generated'])
model = TestModel(config['num_states'], config['num_actions'],
config['models_path_name'])
testsim = TestSimulation(TrafficGen, model, sumo_cmd, config['max_steps'],
config['green_duration'],
config['yellow_duration'], config['num_states'],
config['num_actions'])
queues_episodes = []
waits_episodes = []
simulation_times_episodes = []
episode = 5
simulation_time, avg_queue_length, total_wait_times = testsim.run(
episode, baseline=False)
# while episode < config['total_episodes']:
from testing_simulation import Simulation
from generator import TrafficGenerator
from model import TestModel
from visualization import Visualization
from utils import import_test_configuration, set_sumo, set_test_path
if __name__ == "__main__":
config = import_test_configuration(config_file='testing_settings.ini')
sumo_cmd = set_sumo(config['gui'], config['sumocfg_file_name'],
config['max_steps'])
model_path, plot_path = set_test_path(config['models_path_name'],
config['model_to_test'])
Model = TestModel(input_dim=config['num_states'], model_path=model_path)
TrafficGen = TrafficGenerator(config['max_steps'], config['n_v_generated'])
Visualization = Visualization(plot_path, dpi=96)
Simulation = Simulation(Model, TrafficGen, sumo_cmd, sumo_cmd,
config['max_steps'], config['green_duration'],
config['yellow_duration'], config['num_states'],
config['num_actions'])
print('\n----- Test episode')
simulation_time = Simulation.run(
config['episode_seed']) # run the simulation
print('Simulation time:', simulation_time, 's')
from visualization import Visualization
from model import TestModel
from utils import import_test_configuration, set_sumo, set_test_path
import statistics
if __name__ == "__main__":
config = import_test_configuration(config_file='testing_settings.ini')
sumo_cmd = set_sumo(config['gui'], config['sumocfg_file_name'],
config['max_steps'])
model_path, plot_path = set_test_path(config['models_path_name'],
config['model_to_test'])
Model_A1 = TestModel(input_dim=config['num_states'],
model_path=model_path,
num=1)
Model_A2 = TestModel(input_dim=config['num_states'],
model_path=model_path,
num=2)
TrafficGen = TrafficGenerator(config['max_steps'],
config['n_cars_generated'], "EW")
Visualization = Visualization(plot_path, dpi=96)
Simulation = Simulation(Model_A1, Model_A2, TrafficGen, sumo_cmd,
config['max_steps'], config['green_duration'],
config['yellow_duration'], config['num_cells'],
config['num_states'], config['num_actions'],
help='Visualize training progress')
parser.add_argument(
'--batchsize',
'-b',
default=1,
help='Batch size for training the network (default: %(default)s)')
args = parser.parse_args()
np.random.seed(args.seed)
# loss = Rosenbrock() # optimum: (1,1)
loss = Rosenbrock() # optimum: (3, 0.5)
n = 2
# model = TestModel(n, lr=0.00001, loss=loss, d=np.array([[0.0],[0.0]]))
model = TestModel(n, lr=0.00001, loss=loss, d=np.array([[-1.2], [1]]))
optimizer = None
if args.optimizer == 'ibfgs':
optimizer = InverseBFGS(nparams=model.nparams, gamma=0.0001, eta=0.9)
elif args.optimizer == 'bfgs':
optimizer = BFGS(nparams=model.nparams, gamma=0.0001, eta=0.9)
elif args.optimizer == 'armijo':
optimizer = DescentMethod(nparams=model.nparams, beta=1 / 2, gamma=0.0001)
elif args.optimizer == 'bbv1' or args.optimizer == 'barzilaiborweinv1':
optimizer = BarzilaiBorwein(nparams=model.nparams,
beta=1 / 2,
gamma=0.0001,
strategy='v1')
elif args.optimizer == 'bbv2' or args.optimizer == 'barzilaiborweinv2':
optimizer = BarzilaiBorwein(nparams=model.nparams,
sumo_cmd = set_sumo(config['gui'], config['sumocfg_file_name'],
config['max_steps'])
model_path, plot_path = set_test_path(config['models_path_name'],
config['model_to_test'],
config['scenario_number'])
#SET STATE DIMENSION PARAMETERS
number_of_cells_per_lane = 10
conv_state_shape = (number_of_cells_per_lane, 8, 2)
green_phase_state_shape = 4
elapsed_time_state_shape = 1
state_shape = [
conv_state_shape, green_phase_state_shape, elapsed_time_state_shape
]
Model = TestModel(model_path=model_path, state_shape=state_shape)
TrafficGen = TrafficGenerator(config['max_steps'],
config['penetration_rate'])
Visualization = Visualization(plot_path, dpi=96)
#make non-recurrent model
if Model._recurrent == False:
Simulation = VanillaTestSimulation(Model, TrafficGen, sumo_cmd,
config['max_steps'],
config['green_duration'],
config['yellow_duration'],
config['num_actions'],
config['scenario_number'])
# print("make vanilla simulation")
device = '/GPU:0'
num_samples = 12800
inputs = tf.random.uniform((num_samples, input_steps),
-0.1,
0.1,
dtype=tf.float32,
name="input")
targets = tf.random.uniform((num_samples, intent_size),
-0.1,
0.1,
dtype=tf.float32,
name="output")
with tf.device(device):
model = TestModel(vocab_size, embedding_size, hidden_size, batch_size,
intent_size)
optimizer = tf.keras.optimizers.Adam(learning_rate=1e-3)
for epoch in range(epoch_num):
for step, batch in enumerate(get_batch(batch_size, inputs,
targets)):
encoder_inputs = batch[0]
intent_targets = batch[1]
with tf.GradientTape() as tape:
intent_logits = model(encoder_inputs)
cross_entropy = tf.nn.softmax_cross_entropy_with_logits(
intent_targets, intent_logits)
loss = tf.reduce_mean(cross_entropy)
if step % 50 == 0:
print('epoch:{}, step:{}， intent loss:{:.2f}'.format(
def user(username): item = TestModel(id='1') item.text=str(username) print(item.text) return '<h1>welcome! %s</h1>' % username
import sys
sys.path.append('D:\\PycharmProjects\\FlaskBlog')
from model import TestModel
test = TestModel(testid='1')
test.text='test1'
test.save()
