def test_generate_gated_recurrent_neural_network_model(self):
_net_gen: object = NetworkGenerator(
target=TARGET_TEXT,
predictors=PREDICTORS_TEXT,
output_layer_size=5,
train_data_path=TRAIN_DATA_PATH_TEXT,
test_data_path=TEST_DATA_PATH_TEXT,
validation_data_path=VALIDATION_DATA_PATH_TEXT,
models=['gru'],
sep=',').generate_model()
self.assertTrue(expr=isinstance(_net_gen.model, GRU))
def test_generate_multi_layer_perceptron_model(self):
_net_gen: object = NetworkGenerator(
target=TARGET_TEXT,
predictors=PREDICTORS_TEXT,
output_layer_size=5,
train_data_path=TRAIN_DATA_PATH_TEXT,
test_data_path=TEST_DATA_PATH_TEXT,
validation_data_path=VALIDATION_DATA_PATH_TEXT,
models=['mlp'],
sep=',').generate_model()
self.assertTrue(expr=isinstance(_net_gen.model, MLP))
def test_generate_long_short_term_memory_network_model(self):
_net_gen: object = NetworkGenerator(
target=TARGET_TEXT,
predictors=PREDICTORS_TEXT,
output_layer_size=5,
train_data_path=TRAIN_DATA_PATH_TEXT,
test_data_path=TEST_DATA_PATH_TEXT,
validation_data_path=VALIDATION_DATA_PATH_TEXT,
models=['lstm'],
sep=',').generate_model()
self.assertTrue(expr=isinstance(_net_gen.model, LSTM))
def test_forward(self):
_network_generator: NetworkGenerator = NetworkGenerator(target='label',
predictors=['text'],
output_layer_size=2,
train_data_path=DATA_FILE_PATH.get('train'),
test_data_path=DATA_FILE_PATH.get('test'),
validation_data_path=DATA_FILE_PATH.get('val'),
models=['rcnn']
)
_network_generator.get_vanilla_model()
_network_generator.train()
self.assertTrue(expr=len(_network_generator.fitness.keys()) > 0)
def test_get_vanilla_transformer(self):
_net_gen: object = NetworkGenerator(
target=TARGET_TEXT,
predictors=PREDICTORS_TEXT,
output_layer_size=5,
train_data_path=TRAIN_DATA_PATH_TEXT,
test_data_path=TEST_DATA_PATH_TEXT,
validation_data_path=VALIDATION_DATA_PATH_TEXT,
models=['trans'],
model_name='trans',
sep=',')
_model = _net_gen.get_vanilla_model()
def test_train(self):
_net_gen: object = NetworkGenerator(
target=TARGET_TEXT,
predictors=PREDICTORS_TEXT,
output_layer_size=5,
train_data_path=TRAIN_DATA_PATH_TEXT,
test_data_path=TEST_DATA_PATH_TEXT,
validation_data_path=VALIDATION_DATA_PATH_TEXT,
models=['trans'], #list(NETWORK_TYPE.keys()),
sep=',').generate_model()
_model = _net_gen.generate_model()
_model.train()
self.assertTrue(expr=_model.fitness.get('train') is not None)
def test_generate_params(self):
_net_gen: object = NetworkGenerator(
target=TARGET_TEXT,
predictors=PREDICTORS_TEXT,
output_layer_size=5,
train_data_path=TRAIN_DATA_PATH_TEXT,
test_data_path=TEST_DATA_PATH_TEXT,
validation_data_path=VALIDATION_DATA_PATH_TEXT,
models=list(NETWORK_TYPE.keys())).generate_model()
_model = _net_gen.generate_model()
_mutated_param: dict = copy.deepcopy(_model.model_param_mutated)
_net_gen.generate_params(param_rate=0.1, force_param=None)
self.assertTrue(expr=len(_mutated_param.keys()) < len(
_net_gen.model_param_mutated.keys()))
def test_forward(self):
_predictors: List[str] = ['x1', 'x2', 'x3', 'x4']
_network_generator: NetworkGenerator = NetworkGenerator(target='y',
predictors=_predictors,
output_layer_size=2,
x_train=DATA_SET_MLP[_predictors].values,
y_train=DATA_SET_MLP['y'].values,
x_test=DATA_SET_MLP[_predictors].values,
y_test=DATA_SET_MLP['y'].values,
x_val=DATA_SET_MLP[_predictors].values,
y_val=DATA_SET_MLP['y'].values,
#models=['mlp'],
model_name='mlp',
sequential_type='numeric'
)
_network_generator.generate_model()
_network_generator.train()
self.assertTrue(expr=len(_network_generator.fitness.keys()) > 0)
