def load_data(test=False, seed=2020):
"""Load QM9 dataset
Args:
test (bool): when training DeepHyper, set to False.
seed (int): the random seed used to split the data.
Returns:
If test is True, return training, validation and testing data
and labels, task name and transformer.
If test is False, return training and validation data and labels.
"""
# FIXED PARAMETERS
MAX_ATOM = 10
MAX_EDGE = 17
N_FEAT = 75
E_FEAT = 14
FUNC = load_qm9
FEATURIZER = 'Weave'
SPLIT = 'random'
X_train, y_train, X_valid, y_valid, X_test, y_test, \
tasks, transformers = load_molnet_data(func=FUNC,
featurizer=FEATURIZER,
split=SPLIT,
seed=seed,
MAX_ATOM=MAX_ATOM,
MAX_EDGE=MAX_EDGE,
N_FEAT=N_FEAT,
E_FEAT=E_FEAT)
if test:
return (X_train, y_train), (X_valid, y_valid), (X_test, y_test), \
tasks, transformers
else:
# Shrink the data to its one-tenth
X_train = [X_train[i][::10, ...] for i in range(len(X_train))]
X_valid = [X_valid[i][::10, ...] for i in range(len(X_valid))]
y_train = y_train[::10]
y_valid = y_valid[::10]
return (X_train, y_train), (X_valid, y_valid)
def load_data(test=False, seed=2020):
"""Load ESOL dataset
Args:
test (bool): when training DeepHyper, set to False.
seed (int): the random seed used to split the data.
Returns:
If test is True, return training, validation and testing data
and labels, task name and transformer.
If test is False, return training and validation data and labels.
"""
# FIXED PARAMETERS
MAX_ATOM = 56
MAX_EDGE = 69
N_FEAT = 75
E_FEAT = 14
FUNC = load_delaney
FEATURIZER = 'Weave'
SPLIT = 'random'
X_train, y_train, X_valid, y_valid, X_test, y_test, \
tasks, transformers = load_molnet_data(func=FUNC,
featurizer=FEATURIZER,
split=SPLIT,
seed=seed,
MAX_ATOM=MAX_ATOM,
MAX_EDGE=MAX_EDGE,
N_FEAT=N_FEAT,
E_FEAT=E_FEAT)
if test:
return (X_train, y_train), (X_valid, y_valid), (X_test, y_test), \
tasks, transformers
else:
return (X_train, y_train), (X_valid, y_valid)