def setUp(self):
# TODO cleanup the tests_input directories - not sure if I want by
# default to do the cleanup
utils.ensure_dir(definitions.TEST_FORMULAS_DIR)
utils.ensure_dir(definitions.TEST_SAMPLES_DIR)
for config in os.listdir(definitions.BNN_TEST_CFG):
parser = arg_parser.create_parser()
args = parser.parse_args([
'--results_dir', definitions.ROOT_BNN, 'bnn', '--dataset',
dataset, '--config',
os.path.join(definitions.BNN_TEST_CFG, config), 'encode'
])
print('[{}] {}'.format(self._testMethodName, args))
nn = bnn_dataset.BNN(args, stats)
nn.trained_models_dir = os.path.join(definitions.ROOT_BNN,
nn.filename, 'train')
nn.saved_model = os.path.join(nn.trained_models_dir,
nn.filename + '.pt')
nn.load_model(save=True)
#output_filename = os.path.join(definitions.TEST_FORMULAS_DIR, args.encoder +
# '-' + nn.filename + '-test.dimacs')
#converter = mc.BNNConverter(nn.model_dir, output_filename)
encoder = mc.BNNConverter('best')
formula_fname = encoder.encode([nn])
print('output {}'.format(formula_fname))
def setUp(self):
"""Create data for testing"""
self.parser = create_parser()
args = self.parser.parse_args()
debug = args.debug
self.dim = 3
self.n = 8
self.X = np.array([
[0, 0, 0],
[0, 0, 2],
[0, 1, 0],
[0, 2, 1],
[1, 0, 0],
[1, 0, 1],
[1, 1, 0],
[2, 1, 1],
])
self.feature_names = ['x{}'.format(i) for i in range(self.dim)]
self.index = ['{}'.format(i) for i in range(self.n)]
if debug:
self.log_level = 10
else:
self.log_level = 30
def main():
parser = create_parser()
args = parser.parse_args()
try:
args.func(args)
except TypeError:
parser.print_help()
def setUp(self):
args = create_parser().parse_args()
debug = args.debug
self.seed = args.seed
self.n = args.n
self.dim = args.dim
self.features = [str(i) for i in range(self.dim)]
self.X = norm.rvs(0, 1, size=(self.n, self.dim), random_state=self.seed)
self.B = np.array([-10.1, 2.2, 6.1])
self.y = np.dot(self.X, self.B)
self.y_as_int = np.round(expit(self.y))
self.y_as_string = np.array([str(i) for i in self.y_as_int])
# example dataset for y = B.X
# X = array([[ 1.62434536, -0.61175641, -0.52817175], ... [-0.15065961, -1.40002289, -1.30106608]]) (1000 * 3)
# B = array([-10.1, 2.2, 6.1])
# y = array([ -2.09736000e+01, -1.29850618e+00, -1.73511155e+01, ...]) (1000 * 1)
# features = ['0', '1', '2']
##
# Other output types:
# y_as_int = array[ 0., 0., 0., 0., 1., 1., 0., 0., 0., 1., 1., 1., 1., ...]
# y_as_string = array['0.0', '0.0', '0.0', '0.0', '1.0', '1.0', '0.0', '0.0', '0.0', ... ]
# Another set of input
# sample data
self.sample_x = np.array([[-2, -1], [-1, -1], [-1, -2], [1, 1], [1, 2], [2, 1]])
self.sample_y = np.array([-1, -1, -1, 1, 1, 1])
self.sample_feature_name = [str(i) for i in range(self.sample_x.shape[1])]
if debug:
self.interpreter = Interpretation(log_level='DEBUG')
else:
self.interpreter = Interpretation() # default level is 'WARNING'
self.interpreter.load_data(self.X, feature_names=self.features)
self.regressor = LinearRegression()
self.regressor.fit(self.X, self.y)
self.regressor_predict_fn = InMemoryModel(self.regressor.predict, examples=self.X)
self.classifier = LogisticRegression()
self.classifier.fit(self.X, self.y_as_int)
self.classifier_predict_fn = InMemoryModel(self.classifier.predict, examples=self.X, unique_values=self.classifier.classes_)
self.classifier_predict_proba_fn = InMemoryModel(self.classifier.predict_proba, examples=self.X)
self.string_classifier = LogisticRegression()
self.string_classifier.fit(self.X, self.y_as_string)
self.string_classifier_predict_fn = InMemoryModel(self.string_classifier.predict_proba, examples=self.X)
# Yet another set of input!!
self.sample_x_categorical = np.array([['B', -1], ['A', -1], ['A', -2], ['C', 1], ['C', 2], ['A', 1]])
self.sample_y_categorical = np.array(['A', 'A', 'A', 'B', 'B', 'B'])
self.categorical_feature_names = ['Letters', 'Numbers']
self.categorical_transformer = MultiColumnLabelBinarizer()
self.categorical_transformer.fit(self.sample_x_categorical)
self.sample_x_categorical_transormed = self.categorical_transformer.transform(self.sample_x_categorical)
self.categorical_classifier = LogisticRegression()
self.categorical_classifier.fit(self.sample_x_categorical_transormed, self.sample_y_categorical)
self.categorical_predict_fn = lambda x: self.categorical_classifier.predict_proba(self.categorical_transformer.transform(x))
self.categorical_model = InMemoryModel(self.categorical_predict_fn, examples=self.sample_x_categorical)
def setUp(self):
"""Create data for testing"""
self.parser = create_parser()
args = self.parser.parse_args()
debug = args.debug
self.dim = 3
self.n = 8
self.X = np.array([
[0, 0, 0],
[0, 0, 2],
[0, 1, 0],
[0, 2, 1],
[1, 0, 0],
[1, 0, 1],
[1, 1, 0],
[2, 1, 1],
])
self.feature_names = ['x{}'.format(i) for i in range(self.dim)]
self.index = ['{}'.format(i) for i in range(self.n)]
if debug:
self.log_level = 10
else:
self.log_level = 30
def setUp(self):
args = create_parser().parse_args()
debug = args.debug
self.seed = args.seed
self.n = args.n
self.dim = args.dim
self.features = [str(i) for i in range(self.dim)]
self.X = norm.rvs(0, 1, size=(self.n, self.dim), random_state=self.seed)
self.B = np.array([-10.1, 2.2, 6.1])
self.y = np.dot(self.X, self.B)
self.y_as_int = np.round(expit(self.y))
self.y_as_string = np.array([str(i) for i in self.y_as_int])
# example dataset for y = B.X
# X = array([[ 1.62434536, -0.61175641, -0.52817175], ... [-0.15065961, -1.40002289, -1.30106608]]) (1000 * 3)
# B = array([-10.1, 2.2, 6.1])
# y = array([ -2.09736000e+01, -1.29850618e+00, -1.73511155e+01, ...]) (1000 * 1)
# features = ['0', '1', '2']
##
# Other output types:
# y_as_int = array[ 0., 0., 0., 0., 1., 1., 0., 0., 0., 1., 1., 1., 1., ...]
# y_as_string = array['0.0', '0.0', '0.0', '0.0', '1.0', '1.0', '0.0', '0.0', '0.0', ... ]
# Another set of input
# sample data
self.sample_x = np.array([[-2, -1], [-1, -1], [-1, -2], [1, 1], [1, 2], [2, 1]])
self.sample_y = np.array([-1, -1, -1, 1, 1, 1])
self.sample_feature_name = [str(i) for i in range(self.sample_x.shape[1])]
if debug:
self.interpreter = Interpretation(self.X, feature_names=self.features, log_level='DEBUG')
else:
self.interpreter = Interpretation(self.X, feature_names=self.features) # default level is 'WARNING'
self.regressor = LinearRegression()
self.regressor.fit(self.X, self.y)
self.regressor_predict_fn = InMemoryModel(self.regressor.predict, examples=self.X)
self.classifier = LogisticRegression()
self.classifier.fit(self.X, self.y_as_int)
self.classifier_predict_fn = InMemoryModel(self.classifier.predict, examples=self.X, unique_values=self.classifier.classes_)
self.classifier_predict_proba_fn = InMemoryModel(self.classifier.predict_proba, examples=self.X)
self.string_classifier = LogisticRegression()
self.string_classifier.fit(self.X, self.y_as_string)
self.string_classifier_predict_fn = InMemoryModel(self.string_classifier.predict_proba, examples=self.X)
# Yet another set of input!!
self.sample_x_categorical = np.array([['B', -1], ['A', -1], ['A', -2], ['C', 1], ['C', 2], ['A', 1]])
self.sample_y_categorical = np.array(['A', 'A', 'A', 'B', 'B', 'B'])
self.categorical_feature_names = ['Letters', 'Numbers']
self.categorical_transformer = MultiColumnLabelBinarizer()
self.categorical_transformer.fit(self.sample_x_categorical)
self.sample_x_categorical_transormed = self.categorical_transformer.transform(self.sample_x_categorical)
self.categorical_classifier = LogisticRegression()
self.categorical_classifier.fit(self.sample_x_categorical_transormed, self.sample_y_categorical)
self.categorical_predict_fn = lambda x: self.categorical_classifier.predict_proba(self.categorical_transformer.transform(x))
self.categorical_model = InMemoryModel(self.categorical_predict_fn, examples=self.sample_x_categorical)
def setup_and_parse_args():
parser = arg_parser.create_parser(desc="""
Setup API Gateway in the specified prefix and region. Prefix implies use case
Example usage:
python ./api_create.py --file=<config-file.yaml> # (region us-east-1, developer environment)
""")
return arg_parser.parse_config_from_args(parser)
def setup_and_parse_args():
parser = arg_parser.create_parser(desc="""
Setup Auto scale group.
Example usage:
python ./auto_scale_create.py --file=<config-file.yaml> # (region us-east-1, developer environment)
""")
return arg_parser.parse_config_from_args(parser)
def setup_and_parse_args():
parser = arg_parser.create_parser(desc="""
Setup EC2 in the specified region. Environment implies both region and its use case
Example usage:
python ./ec2_create.py --file=<config-file.yaml> # (region us-east-1, developer environment)
""")
return arg_parser.parse_config_from_args(parser)
def test_all_bnn_configs(self):
pool = Pool(processes=20)
test_args = []
for config in os.listdir(definitions.BNN_TEST_CFG):
parser = arg_parser.create_parser()
args = parser.parse_args([
'--results_dir', definitions.ROOT_BNN, 'bnn', '--dataset',
dataset, '--config',
os.path.join(definitions.BNN_TEST_CFG, config), 'encode'
])
test_args.append((args, stats))
res = pool.map(run_bnn_cfg, test_args)
for r in res:
self.assertEqual(r[1], 1.0)
def main():
f_name = None
test_passed = False
parser = create_parser()
args = parser.parse_args()
hook_enabled = False
if args.url:
f_name = download_file(args.url[0])
elif args.filename:
f_name = args.filename[0]
else:
print('No valid archive was provided, aborting...')
exit(-1)
cont = unpack_file(f_name=f_name)
if args.vmname:
if args.planname:
if args.hook:
hook_enabled = True
print('<<<---Looking for hook--->>>')
plan = Plan(args.planname[0], cont)
plan.first_init()
for current_vm in args.vmname:
plan.vm_list.append(Vm(current_vm, cont, hook_enabled))
assert len(plan.vm_list) == len(args.vmname)
plan.validate()
if not plan.error_state:
test_passed = True
else:
print('No plan was provided, only VM and DV will be inspected')
vm = Vm(args.vmname[0], cont)
vm.validate()
if not vm.error_state:
test_passed = True
else:
print('No VM names were provided, aborting...')
exit(-1)
if test_passed:
print('PASSED')
else:
print('FAILED')
def test_unsat_diff_cpp(self):
pool = Pool(processes=20)
test_args = []
for config in os.listdir(definitions.BNN_TEST_CFG):
parser = arg_parser.create_parser()
# these are some dummy args just so we can create the BNNmnist and
# initialize the correct paths of model_dir and enc_out_dir
args = parser.parse_args(['--results_dir', definitions.ROOT_BNN,
'bnn-mnist', '--config',
os.path.join(definitions.BNN_TEST_CFG,
config), 'test_enc'])
test_args.append((args, stats))
res = pool.map(run_unsat_call_cpp, test_args)
for sol in res:
if sol == z3.unsat:
print('[{}] PASS {}'.format(self._testMethodName, args.config))
self.assertEqual(sol, z3.unsat)
def setUp(self):
self.parser = arg_parser.create_parser()
def main():
parser = create_parser()
args = parser.parse_args()
args.func(args)
#!/usr/bin/env python
import arg_parser
import stats
parser = arg_parser.create_parser()
args = parser.parse_args()
args.func(args)
def main():
pass
if __name__ == "__main__":
main()
def main():
parser = create_parser()
args = parser.parse_args()
fusionfusion_main(args)
