def test_triplet_fit():
(x, y), (_, _) = tf.keras.datasets.mnist.load_data()
train = batch.BatchDataset(x[:10], y[:10], input_shape=(10, 784))
new_base = mlp.MLP()
new_siamese = triplet.TripletSiamese(new_base, loss="hard")
new_siamese.compile(optimizer=tf.optimizers.Adam(learning_rate=0.001))
new_siamese = triplet.TripletSiamese(new_base, loss="semi-hard")
new_siamese.compile(optimizer=tf.optimizers.Adam(learning_rate=0.001))
new_siamese.fit(train.batches, epochs=1)
def test_triplet_margin_setter():
new_base = mlp.Base()
new_siamese = triplet.TripletSiamese(new_base)
try:
new_siamese.margin = -1
except:
new_siamese.margin = 1.0
assert new_siamese.margin == 1.0
def test_triplet_distance_setter():
new_base = mlp.Base()
new_siamese = triplet.TripletSiamese(new_base)
try:
new_siamese.distance = "a"
except:
new_siamese.distance == "L2"
assert new_siamese.distance == "squared-L2"
def test_triplet_soft_setter():
new_base = mlp.Base()
new_siamese = triplet.TripletSiamese(new_base)
try:
new_siamese.soft = -1
except:
new_siamese.soft = False
assert new_siamese.soft is False
def test_triplet_loss_type_setter():
new_base = mlp.Base()
new_siamese = triplet.TripletSiamese(new_base)
try:
new_siamese.loss_type = "a"
except:
new_siamese.loss_type = "hard"
assert new_siamese.loss_type == "hard"
def test_triplet_step():
(x, y), (_, _) = tf.keras.datasets.mnist.load_data()
new_base = mlp.MLP()
new_siamese = triplet.TripletSiamese(new_base)
new_siamese.compile(optimizer=tf.optimizers.Adam(learning_rate=0.001))
x = tf.ones((10, 784))
y = tf.zeros(10)
new_siamese.step(x, y)
def test_triplet_predict():
(x, y), (_, _) = tf.keras.datasets.mnist.load_data()
train = batch.BatchDataset(x[:10], y[:10], input_shape=(10, 784))
new_base = mlp.MLP()
new_siamese = triplet.TripletSiamese(new_base)
new_siamese.compile(optimizer=tf.optimizers.Adam(learning_rate=0.001))
new_siamese.fit(train.batches, epochs=1)
x1 = tf.ones((1, 784))
x2 = tf.ones((1, 784))
new_siamese.distance = "L1"
new_siamese.predict(x1, x2)
new_siamese.distance = "L2"
new_siamese.predict(x1, x2)
new_siamese.distance = "angular"
new_siamese.predict(x1, x2)
def test_triplet_loss_type():
new_base = mlp.Base()
new_siamese = triplet.TripletSiamese(new_base)
assert new_siamese.loss_type == "hard"
def test_triplet_distance():
new_base = mlp.Base()
new_siamese = triplet.TripletSiamese(new_base)
assert new_siamese.distance == "squared-L2"
def test_triplet_margin():
new_base = mlp.Base()
new_siamese = triplet.TripletSiamese(new_base)
assert new_siamese.margin == 1.0
def test_triplet_soft():
new_base = mlp.Base()
new_siamese = triplet.TripletSiamese(new_base)
assert new_siamese.soft is False