def test_mp_metallicity_loader():
current_dir = os.path.dirname(os.path.abspath(__file__))
tasks, datasets, transformers = load_mp_metallicity(
reload=False,
data_dir=current_dir,
featurizer_kwargs={'max_atoms': 8},
splitter_kwargs={
'seed': 42,
'frac_train': 0.6,
'frac_valid': 0.2,
'frac_test': 0.2
})
assert tasks[0] == 'is_metal'
assert datasets[0].X.shape == (3, 8)
assert datasets[1].X.shape == (1, 8)
assert datasets[2].X.shape == (1, 8)
assert np.allclose(datasets[0].X[0], [
0.80428488, -0.70720997, 1.29101261, 0.61631094, 0.84184489,
-0.28273997, -1.10252907, -1.23500371
],
atol=0.01)
if os.path.exists(os.path.join(current_dir, 'mp_is_metal.json')):
os.remove(os.path.join(current_dir, 'mp_is_metal.json'))
def test_cgcnn_reload():
# load datasets
current_dir = path.dirname(path.abspath(__file__))
config = {
"reload": False,
"featurizer": CGCNNFeaturizer(),
# disable transformer
"transformers": [],
"data_dir": path.join(current_dir, "assets")
}
tasks, datasets, transformers = load_mp_metallicity(**config)
train, valid, test = datasets
n_tasks = len(tasks)
n_classes = 2
model_dir = tempfile.mkdtemp()
model = CGCNNModel(
n_tasks=n_tasks,
n_classes=n_classes,
mode='classification',
model_dir=model_dir,
batch_size=4,
learning_rate=0.001)
# check train
model.fit(train, nb_epoch=20)
# check predict shape
valid_preds = model.predict_on_batch(valid.X)
assert valid_preds.shape == (2, n_classes)
test_preds = model.predict(test)
assert test_preds.shape == (3, n_classes)
# check overfit
classification_metric = Metric(roc_auc_score, n_tasks=n_tasks)
scores = model.evaluate(
train, [classification_metric], transformers, n_classes=n_classes)
assert scores[classification_metric.name] > 0.8
# reload
reloaded_model = CGCNNModel(
n_tasks=n_tasks,
n_classes=n_classes,
mode='classification',
model_dir=model_dir,
batch_size=4,
learning_rate=0.001)
reloaded_model.restore()
original_pred = model.predict(test)
reload_pred = reloaded_model.predict(test)
assert np.all(original_pred == reload_pred)
if path.exists(path.join(current_dir, 'mp_is_metal.json')):
remove(path.join(current_dir, 'mp_is_metal.json'))
def test_cgcnn():
# regression test
# load datasets
current_dir = path.dirname(path.abspath(__file__))
config = {
"reload": False,
"featurizer": CGCNNFeaturizer,
# disable transformer
"transformers": [],
"data_dir": current_dir
}
tasks, datasets, transformers = load_perovskite(**config)
train, valid, test = datasets
n_tasks = len(tasks)
model = CGCNNModel(n_tasks=n_tasks,
mode='regression',
batch_size=4,
learning_rate=0.001)
# check train
model.fit(train, nb_epoch=20)
# check predict shape
valid_preds = model.predict_on_batch(valid.X)
assert valid_preds.shape == (2, n_tasks)
test_preds = model.predict(test)
assert test_preds.shape == (3, n_tasks)
# check overfit
regression_metric = Metric(mae_score, n_tasks=n_tasks)
scores = model.evaluate(train, [regression_metric], transformers)
assert scores[regression_metric.name] < 0.6
# classification test
tasks, datasets, transformers = load_mp_metallicity(**config)
train, valid, test = datasets
# load datasets
n_tasks = len(tasks)
n_classes = 2
model = CGCNNModel(n_tasks=n_tasks,
n_classes=n_classes,
mode='classification',
batch_size=4,
learning_rate=0.001)
# check train
model.fit(train, nb_epoch=20)
# check predict shape
valid_preds = model.predict_on_batch(valid.X)
assert valid_preds.shape == (2, n_classes)
test_preds = model.predict(test)
assert test_preds.shape == (3, n_classes)
# check overfit
classification_metric = Metric(roc_auc_score, n_tasks=n_tasks)
scores = model.evaluate(train, [classification_metric],
transformers,
n_classes=n_classes)
assert scores[classification_metric.name] > 0.8
# TODO: Multi task classification test
if path.exists(path.join(current_dir, 'perovskite.json')):
remove(path.join(current_dir, 'perovskite.json'))
if path.exists(path.join(current_dir, 'mp_is_metal.json')):
remove(path.join(current_dir, 'mp_is_metal.json'))