from preprocess import get_test_train_data
from save_load import get_last_file_number
from save_load import load
from model import accuracy, loss
from keras import backend
import sys, os
import tensorflow as tf
from utils import visualize_layers
if __name__ == '__main__':
file = "F:\Projects\python\self_driving_game\data\dataset_mini.pz"
if len(sys.argv)<=1: count = None
else: count = int(sys.argv[1]);
# Load model
exp_folder = 'exp_' + '{0:03d}'.format(get_last_file_number(prefix='exp_', suffix=''))
model = load(count, path=exp_folder)
# Visualize model
x_train, x_test, y_train, y_test = get_test_train_data(file, 10, tanh=True)
visualization_folder = exp_folder + '/visualization'
if not os.path.exists(visualization_folder): os.makedirs(visualization_folder)
visualize_layers(model, x_test, path=visualization_folder)
print('Visualization done...!')
file = os.environ['DATA_DIR'] + "/dataset_75p_gray.pz"
x_train, x_test, y_train, y_test = get_test_train_data(file,
80000,
tanh=False)
# x_train, x_test, y_train, y_test = get_test_train_data(file, 1000, tanh=False)
learning_rate = 0.0005
models = ['relu_with_scaled_sigmoid']
drop_rates = [0.1]
for i in range(len(models)):
for j in range(len(drop_rates)):
# Print output to file
outfolder = 'exp_' + '{0:03d}'.format(
get_last_file_number(prefix='exp_', suffix='') + 1)
os.makedirs(outfolder)
outfile = outfolder + '/' + 'train_' + '{0:03d}'.format(
get_last_file_number(path=outfolder) + 1) + '.log'
print('Printing to logfile at', outfile)
sys.stdout = open(outfile, 'w+')
print(
'Title:',
'{}_adam_{}_dropout_rate_{}'.format(models[i], learning_rate,
drop_rates[j]), '\n\n')
if models[i] == 'relu':
model = relu_model(learning_rate=learning_rate)
elif models[i] == 'relu_with_scaled_sigmoid':
model = relu_with_scaled_sigmoid_model(
y_pred = model.predict(x_test)
y_pred = backend.cast(y_pred, 'float32')
current_accuracy = accuracy(y_test, y_pred)
current_loss = loss(y_test, y_pred)
return [current_loss, current_accuracy]
if __name__ == '__main__':
# file = "F:\Projects\python\self_driving_game\data\dataset_mini.pz"
file = "data\dataset_mini.pz"
if len(sys.argv) <= 1: count = None
else: count = int(sys.argv[1])
# Load model
exp_folder = 'experiments/comparison/activation_and_lrs/exp_' + '{0:03d}'.format(
get_last_file_number(prefix='exp_', suffix=''))
model = load(count, path=exp_folder)
# Evaluate model
x_train, x_test, y_train, y_test = get_test_train_data(file,
10000,
tanh=True)
print('x_test shape', x_test.shape)
scores = evaluate_model(model, x_test, y_test)
# Print scores
print('\n\n')
print("Loss: ", backend.get_value(scores[0]))
print("Accuracy: ", backend.get_value(scores[1]) * 100, "%")
import sys, os
import tensorflow as tf
def evaluate_model(model, x_test, y_test):
y_pred = model.predict(x_test)
y_pred = backend.cast(y_pred, 'float32')
current_accuracy = accuracy(y_test, y_pred)
current_loss = loss(y_test, y_pred)
return [current_loss, current_accuracy]
if __name__ == '__main__':
# file = "F:\Projects\python\self_driving_game\data\dataset_mini.pz"
file = os.environ['DATA_DIR']+ "/dataset_75p_gray.pz"
if len(sys.argv)<=1: count = None
else: count = int(sys.argv[1]);
# Load model
exp_folder = 'experiments/comparison/activation_and_lrs/exp_' + '{0:03d}'.format(get_last_file_number(prefix='exp_', suffix=''))
model = load(count, path=exp_folder)
# Evaluate model
x_train, x_test, y_train, y_test = get_test_train_data(file, 10000, tanh=True)
print('x_test shape', x_test.shape)
scores = evaluate_model(model, x_test, y_test)
# Print scores
print('\n\n')
print("Loss: ", backend.get_value(scores[0]))
print("Accuracy: ", backend.get_value(scores[1])*100, "%")
