def setUpClass(cls):
cls.tf_data = tf.Variable([1.0, 2.0, 4.0])
cls.tf_output = [tf.constant([2, 4, 8])]
cls.torch_data = torch.tensor([1.0, 2.0, 4.0], requires_grad=True)
cls.torch_output = [torch.tensor([2, 4, 8], dtype=torch.float32)]
cls.tf_model = one_layer_tf_model()
cls.torch_model = OneLayerTorchModel()
def instantiate_system():
system = sample_system_object()
submodel = one_layer_tf_model()
model = fe.build(model_fn=lambda: test_model(submodel),
optimizer_fn='adam',
model_name='tf')
model2 = fe.build(model_fn=lambda: test_model(submodel),
optimizer_fn='adam',
model_name='tf2')
system.network = fe.Network(ops=[
ModelOp(model=model, inputs="x_out", outputs="y_pred"),
ModelOp(model=model2, inputs="x_out", outputs="y_pred2"),
])
return system
def setUpClass(cls):
cls.tf_model = one_layer_tf_model()
cls.torch_model = OneLayerTorchModel()
def test_feed_forward_tf(self):
model = one_layer_tf_model()
x = tf.constant([[1.0, 1.0, 1.0], [1.0, -1.0, -0.5]])
obj1 = fe.backend.feed_forward(model, x)
obj2 = tf.constant([[6.0], [-2.5]])
self.assertTrue(is_equal(obj1, obj2))