def test_conv_deeper_stub():
model = get_conv_model()
graph = Graph(to_stub_model(model))
layer_num = graph.n_layers
graph.to_conv_deeper_model(3, 3)
assert graph.n_layers == layer_num + 3
def test_conv_deeper_stub():
model = get_conv_model()
graph = Graph(model, False)
layer_num = graph.n_layers
graph.to_conv_deeper_model(6, 3)
assert graph.n_layers == layer_num + 4
def test_wider_conv():
model = to_stub_model(get_conv_model(), True)
assert isinstance(wider_pre_conv(model.layers[2], 3), StubConv)
assert isinstance(wider_bn(model.layers[3], 3, 3, 3),
StubBatchNormalization)
assert isinstance(wider_next_conv(model.layers[6], 3, 3, 3), StubConv)
def test_conv_deeper():
model = get_conv_model()
graph = Graph(model)
new_model = graph.to_conv_deeper_model(model.layers[4], 3)
input_data = get_conv_data()
output1 = model.predict_on_batch(input_data).flatten()
output2 = new_model.predict_on_batch(input_data).flatten()
assert abs(np.sum(output1 - output2)) < 0.2
def test_skip_add():
model = get_conv_model()
graph = Graph(model)
new_model = graph.to_add_skip_model(model.layers[1], model.layers[4])
input_data = get_conv_data()
output1 = model.predict_on_batch(input_data).flatten()
output2 = new_model.predict_on_batch(input_data).flatten()
assert np.array_equal(output1, output2)
def test_conv_deeper():
model = get_conv_model()
graph = Graph(model, True)
graph.to_conv_deeper_model(6, 3)
new_model = graph.produce_model()
input_data = get_conv_data()
output1 = model.predict_on_batch(input_data).flatten()
output2 = new_model.predict_on_batch(input_data).flatten()
assert np.sum(np.abs(output1 - output2)) < 4e-1
def test_deeper_conv_block():
model = get_conv_model()
layers = deeper_conv_block(model.layers[1], 3)
output_tensor = layers[0](model.outputs[0])
output_tensor = layers[1](output_tensor)
output_tensor = layers[2](output_tensor)
new_model = Model(inputs=model.inputs, outputs=output_tensor)
input_data = get_conv_data()
output1 = model.predict_on_batch(input_data).flatten()
output2 = new_model.predict_on_batch(input_data).flatten()
assert np.sum(np.abs(output1 - output2)) < 1e-1
def test_conv_deeper():
model = get_conv_model()
graph = NetworkMorphismGraph(model)
graph.to_conv_deeper_model(graph.layer_to_id[model.layers[4]], 3)
new_model = graph.produce_model()
input_data = get_conv_data()
output1 = model.predict_on_batch(input_data).flatten()
output2 = new_model.predict_on_batch(input_data).flatten()
assert np.sum(np.abs(output1 - output2)) < 1e-1
def test_wider_conv():
model = get_conv_model()
layer1 = wider_pre_conv(model.layers[1], 3)
layer2 = wider_bn(model.layers[2], 3, 3, 3)
layer3 = wider_next_conv(model.layers[4], 3, 3, 3)
input_tensor = Input(shape=(5, 5, 3))
output_tensor = layer1(input_tensor)
output_tensor = layer2(output_tensor)
output_tensor = Activation('relu')(output_tensor)
output_tensor = layer3(output_tensor)
output_tensor = BatchNormalization()(output_tensor)
output_tensor = Activation('relu')(output_tensor)
model2 = Model(inputs=input_tensor, outputs=output_tensor)
random_input = get_conv_data()
output1 = model.predict_on_batch(random_input)
output2 = model2.predict_on_batch(random_input)
assert np.sum(np.abs(output1.flatten() - output2.flatten())) < 1e-1
def test_graph():
graph = Graph(get_conv_model())
assert graph.n_nodes == 13
def test_wider_bn():
bn_layer = get_conv_model().layers[2]
new_bn_layer = wider_bn(bn_layer, 1, 3, 4)
assert new_bn_layer.get_weights()[0].shape[0] == 7
def test_deeper_conv_block():
model = to_stub_model(get_conv_model(), True)
layers = deeper_conv_block(model.layers[2], 3)
assert len(layers) == constant.CONV_BLOCK_DISTANCE + 1