def create_nam_model(x_train,
dropout,
feature_dropout=0.0,
num_basis_functions=1000,
units_multiplier=2,
activation='exu',
name_scope='model',
shallow=True,
trainable=True):
"""Create the NAM model."""
num_unique_vals = [
len(np.unique(x_train[:, i])) for i in range(x_train.shape[1])
]
num_units = [
min(num_basis_functions, i * units_multiplier) for i in num_unique_vals
]
num_inputs = x_train.shape[-1]
nn_model = models.NAM(num_inputs=num_inputs,
num_units=num_units,
dropout=np.float32(dropout),
feature_dropout=np.float32(feature_dropout),
activation=activation,
shallow=shallow,
trainable=trainable,
name_scope=name_scope)
return nn_model
def test_model(self, architecture):
"""Test whether a model with specified architecture can be run."""
x = np.random.rand(5, 10).astype('float32')
sess = tf.InteractiveSession()
if architecture == 'exu_nam':
model = models.NAM(num_inputs=x.shape[1],
num_units=1024,
shallow=True,
activation='exu')
elif architecture == 'relu_nam':
model = models.NAM(num_inputs=x.shape[1],
num_units=64,
shallow=False,
activation='relu')
elif architecture == 'dnn':
model = models.DNN()
else:
raise ValueError('Architecture {} not found'.format(architecture))
out_op = model(x)
sess.run(tf.global_variables_initializer())
self.assertIsInstance(sess.run(out_op), np.ndarray)
sess.close()