IntenseWebs/openvino
3
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Added support of shapes and types from original FW in ov.convert_model() (#20009)

Browse Source 
* Added support of shapes and types from paddle, torch and tf.

* Removed changes from requirements.

* Corrected test.

* Moved helper methods to utils.

* Separated tests by frameworks.

* Removed changes from complex_params test.
This commit is contained in:
Anastasiia Pnevskaia
2023-09-26 17:41:01 +04:00
committed by GitHub
parent c3565e3eac
commit 2bbfe7b44d
7 changed files with 344 additions and 60 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/layer_tests/ovc_python_api_tests/test_paddle.py
+32
View File
@@ -101,3 +101,35 @@ class TestMoConvertPaddle(CommonMOConvertTest):
if mo_params is not None:
test_params.update(mo_params)
self._test_by_ref_graph(temp_dir, test_params, graph_ref, compare_tensor_names=False)
class TestPaddleConversionParams(CommonMOConvertTest):
paddle_is_imported = False
try:
import paddle
paddle_is_imported = True
except ImportError:
pass
test_data = [
{'params_test': {'input': paddle.shape(paddle.to_tensor(np.random.rand(2, 3, 4)))},
'fw_model': make_pd_hapi_graph_model([1, 2]),
'ref_model': make_ref_graph_model([2, 3, 4])},
{'params_test': {'input': paddle.to_tensor(np.random.rand(5, 6)).shape},
'fw_model': make_pd_hapi_graph_model([1, 2, 3]),
'ref_model': make_ref_graph_model([5, 6])},
{'params_test': {'input': (paddle.to_tensor(np.random.rand(4, 2, 7)).shape, paddle.int32)},
'fw_model': make_pd_hapi_graph_model([2, 3]),
'ref_model': make_ref_graph_model([4, 2, 7], np.int32)},
] if paddle_is_imported else []
@pytest.mark.parametrize("params", test_data)
@pytest.mark.nightly
def test_conversion_params(self, params, ie_device, precision, ir_version,
temp_dir, use_new_frontend, use_old_api):
fw_model = params['fw_model']
test_params = params['params_test']
ref_model = params['ref_model']
test_params.update({'input_model': fw_model})
self._test_by_ref_graph(temp_dir, test_params, ref_model, compare_tensor_names=False)
tests/layer_tests/ovc_python_api_tests/test_pytorch.py
+72
View File
@@ -1145,3 +1145,75 @@ class ConvertRaises(unittest.TestCase):
with self.assertRaisesRegex(Exception, ".*Cannot recognize input model.*"):
with tempfile.NamedTemporaryFile() as tmpfile:
convert_model(tmpfile.name)
def create_model_three_inputs():
from torch import nn
class NeuralNetwork(nn.Module):
def __init__(self):
super(NeuralNetwork, self).__init__()
self.linear_relu_stack = nn.Sequential(
nn.ReLU(),
nn.Sigmoid(),
)
def forward(self, x, y, z):
out = self.linear_relu_stack(x + y + z),
return out
return NeuralNetwork()
def make_ref_model_three_inputs(shape, dtype=np.float32):
x = ov.opset8.parameter(PartialShape(
shape), name="x", dtype=dtype)
y = ov.opset8.parameter(PartialShape(
shape), name="y", dtype=dtype)
z = ov.opset8.parameter(PartialShape(
shape), name="z", dtype=dtype)
add1 = ov.opset8.add(x, y)
add2 = ov.opset8.add(add1, z)
relu = ov.opset8.relu(add2)
if dtype not in [np.float32, Type.dynamic]:
relu = ov.opset8.convert(relu, np.float32)
sigm = ov.opset8.sigmoid(relu)
parameter_list = [x, y, z]
model = Model([sigm], parameter_list, "test")
return model
class TestPytorchConversionParams(CommonMOConvertTest):
test_data = [
{'params_test': {'input': [(torch.Size([2, 3, 4]), torch.float32),
(torch.empty(2, 3, 4).size(), torch.float32),
(torch.empty(2, 3, 4).shape, torch.float32)]},
'fw_model': create_model_three_inputs(),
'ref_model': make_ref_model_three_inputs([2,3,4], np.float32)},
{'params_test': {'input': [(torch.Size([5, 2]), torch.int32),
(torch.empty(5, 2).size(), torch.int32),
(torch.empty(5, 2).shape, torch.int32)]},
'fw_model': create_model_three_inputs(),
'ref_model': make_ref_model_three_inputs([5, 2], np.int32)},
{'params_test': {'input': [(torch.Size([1, 3, 5]), torch.float32)]},
'fw_model': make_pt_model_one_input(),
'ref_model': make_ref_pt_model_one_input([1, 3, 5], np.float32)},
{'params_test': {'input': [(torch.empty(7, 3).size(), torch.int32)]},
'fw_model': make_pt_model_one_input(),
'ref_model': make_ref_pt_model_one_input([7, 3], np.int32)},
]
@pytest.mark.parametrize("params", test_data)
@pytest.mark.nightly
def test_conversion_params(self, params, ie_device, precision, ir_version,
temp_dir, use_new_frontend, use_old_api):
fw_model = params['fw_model']
test_params = params['params_test']
ref_model = params['ref_model']
test_params.update({'input_model': fw_model})
self._test_by_ref_graph(temp_dir, test_params, ref_model, compare_tensor_names=False)
tests/layer_tests/ovc_python_api_tests/test_tf.py
+98 -39
View File
@@ -10,11 +10,10 @@ from openvino.runtime import PartialShape, Model, Dimension
from common.mo_convert_test_class import CommonMOConvertTest
from common.layer_test_class import CommonLayerTest
import tensorflow as tf
def create_tf_graph_def(tmp_dir):
import tensorflow as tf
tf.compat.v1.reset_default_graph()
with tf.compat.v1.Session() as sess:
@@ -41,8 +40,6 @@ def create_tf_graph_def(tmp_dir):
def create_keras_model(temp_dir):
import tensorflow as tf
tf.keras.backend.clear_session()
tf.compat.v1.reset_default_graph()
@@ -69,8 +66,6 @@ def create_keras_model(temp_dir):
def create_tf1_wrap_function(tmp_dir):
import tensorflow as tf
def f(x, y):
return tf.nn.sigmoid(tf.nn.relu(x + y))
@@ -91,7 +86,6 @@ def create_tf1_wrap_function(tmp_dir):
def create_tf_session(tmp_dir):
import tensorflow as tf
from tensorflow.python.eager.context import graph_mode
with graph_mode():
@@ -119,8 +113,6 @@ def create_tf_session(tmp_dir):
def create_tf_module(tmp_dir):
import tensorflow as tf
class Net(tf.Module):
def __init__(self, name=None):
super(Net, self).__init__(name=name)
@@ -143,8 +135,6 @@ def create_tf_module(tmp_dir):
def create_tf_module_dynamic(tmp_dir):
import tensorflow as tf
class Net(tf.Module):
def __init__(self, name=None):
super(Net, self).__init__(name=name)
@@ -169,7 +159,6 @@ def create_tf_module_dynamic(tmp_dir):
def create_keras_layer(tmp_dir):
import tensorflow as tf
class LayerModel(tf.keras.layers.Layer):
def __init__(self):
@@ -193,7 +182,6 @@ def create_keras_layer(tmp_dir):
def create_keras_layer_dynamic(tmp_dir):
import tensorflow as tf
class LayerModel(tf.keras.layers.Layer):
def __init__(self):
@@ -219,8 +207,6 @@ def create_keras_layer_dynamic(tmp_dir):
def create_tf_checkpoint(tmp_dir):
import tensorflow as tf
input_names = ["Input1", "Input2"]
input_shape = [1, 2, 3]
@@ -245,8 +231,6 @@ def create_tf_checkpoint(tmp_dir):
def create_tf_function(temp_dir):
import tensorflow as tf
@tf.function(
input_signature=[tf.TensorSpec(shape=[1, 2, 3], dtype=tf.float32),
tf.TensorSpec(shape=[1, 2, 3], dtype=tf.float32)])
@@ -268,8 +252,6 @@ def create_tf_function(temp_dir):
def create_tf_graph(temp_dir):
import tensorflow as tf
tf.compat.v1.reset_default_graph()
with tf.compat.v1.Session() as sess:
@@ -296,8 +278,6 @@ def create_tf_graph(temp_dir):
def create_tf_saved_model_dir(temp_dir):
import tensorflow as tf
input_names = ["Input1", "Input2"]
input_shape = [1, 2, 3]
@@ -322,7 +302,6 @@ def create_tf_saved_model_dir(temp_dir):
def create_tf_stateful_partioned_call_net(temp_dir):
import tensorflow as tf
tf.compat.v1.reset_default_graph()
data_shape = [1, 1, 10, 10]
@@ -359,7 +338,6 @@ def create_tf_stateful_partioned_call_net(temp_dir):
def create_keras_layer_input_list():
import tensorflow as tf
class LayerModel(tf.keras.layers.Layer):
def __init__(self):
@@ -386,7 +364,6 @@ def create_keras_layer_input_list():
def create_keras_layer_input_list_one_inp():
import tensorflow as tf
class LayerModel(tf.keras.layers.Layer):
def __init__(self):
@@ -408,7 +385,6 @@ def create_keras_layer_input_list_one_inp():
def create_keras_layer_input_dict():
import tensorflow as tf
class LayerModel(tf.keras.layers.Layer):
def __init__(self):
@@ -434,7 +410,6 @@ def create_keras_layer_input_dict():
def create_keras_layer_input_dict_one_inp():
import tensorflow as tf
class LayerModel(tf.keras.layers.Layer):
def __init__(self):
@@ -522,7 +497,6 @@ def create_keras_layer_with_input_shapes_case4(tmp_dir):
def create_keras_layer_with_tf_function_call(tmp_dir):
import tensorflow as tf
class LayerModel(tf.Module):
def __init__(self):
super(LayerModel, self).__init__()
@@ -538,7 +512,6 @@ def create_keras_layer_with_tf_function_call(tmp_dir):
def create_keras_layer_with_tf_function_call_default_compressed_to_fp16(tmp_dir):
import tensorflow as tf
class LayerModel(tf.Module):
def __init__(self):
super(LayerModel, self).__init__()
@@ -554,7 +527,6 @@ def create_keras_layer_with_tf_function_call_default_compressed_to_fp16(tmp_dir)
def create_keras_layer_with_tf_function_call_no_signature(tmp_dir):
import tensorflow as tf
class LayerModel(tf.Module):
def __init__(self):
super(LayerModel, self).__init__()
@@ -572,7 +544,6 @@ def create_keras_layer_with_tf_function_call_no_signature(tmp_dir):
def create_keras_layer_with_tf_function_call_no_signature_single_input(tmp_dir):
import tensorflow as tf
class LayerModel(tf.Module):
def __init__(self):
super(LayerModel, self).__init__()
@@ -590,7 +561,6 @@ def create_keras_layer_with_tf_function_call_no_signature_single_input(tmp_dir):
def create_keras_layer_with_string_tensor(tmp_dir):
import tensorflow as tf
class LayerModel(tf.Module):
def __init__(self):
super(LayerModel, self).__init__()
@@ -612,6 +582,69 @@ def create_keras_layer_with_string_tensor(tmp_dir):
return model, model_ref, {}
def create_tf_model_three_inputs(shape=[1, 2, 3, 4], type=tf.float32):
tf.compat.v1.reset_default_graph()
with tf.compat.v1.Session() as sess:
inp1 = tf.compat.v1.placeholder(type, shape, 'Input1')
inp2 = tf.compat.v1.placeholder(type, shape, 'Input2')
inp3 = tf.compat.v1.placeholder(type, shape, 'Input3')
relu1 = tf.nn.relu(inp1, name='Relu1')
relu2 = tf.nn.relu(inp2, name='Relu2')
relu3 = tf.nn.relu(inp3, name='Relu3')
add = relu1 + relu2 + relu3
tf.compat.v1.global_variables_initializer()
tf_net = sess.graph
return tf_net
def create_ref_model_three_inputs(shape=[1, 2, 3, 4], dtype=np.float32):
inp1 = ov.opset8.parameter(PartialShape(
shape), name="Input1", dtype=dtype)
inp2 = ov.opset8.parameter(PartialShape(
shape), name="Input2", dtype=dtype)
inp3 = ov.opset8.parameter(PartialShape(
shape), name="Input3", dtype=dtype)
relu1 = ov.opset8.relu(inp1)
relu2 = ov.opset8.relu(inp2)
relu3 = ov.opset8.relu(inp3)
add1 = ov.opset8.add(relu1, relu2)
add2 = ov.opset8.add(add1, relu3)
parameter_list = [inp1, inp2, inp3]
model = Model([add2], parameter_list, "test")
return model
def create_tf_model_single_input(shape=[1, 2, 3, 4], type=tf.float32):
tf.compat.v1.reset_default_graph()
with tf.compat.v1.Session() as sess:
inp = tf.compat.v1.placeholder(type, shape, 'Input')
relu = tf.nn.relu(inp, name='Relu')
output = tf.nn.sigmoid(relu, name='Sigmoid')
tf.compat.v1.global_variables_initializer()
tf_net = sess.graph
return tf_net
def create_ref_model_single_input(shape=[1, 2, 3, 4], dtype=np.float32):
inp = ov.opset8.parameter(PartialShape(
shape), name="Input", dtype=dtype)
relu = ov.opset8.relu(inp)
sigm = ov.opset8.sigmoid(relu)
parameter_list = [inp]
model = Model([sigm], parameter_list, "test")
return model
class TestMoConvertTF(CommonMOConvertTest):
test_data = [
# TF2
@@ -667,7 +700,6 @@ class TestMoConvertTF(CommonMOConvertTest):
self._test_by_ref_graph(temp_dir, test_params, graph_ref, compare_tensor_names=False)
def test_zero_copy(self, ie_device, precision, ir_version, temp_dir):
import tensorflow as tf
from openvino.tools.ovc import convert_model
from openvino.runtime import compile_model
class LayerModel(tf.Module):
@@ -716,7 +748,6 @@ class TestMoConvertTF(CommonMOConvertTest):
assert np.array_equal(ov_infer2['Identity:0'], [ 0., 8., 16.])
def test_turn_off_sharing(self, ie_device, precision, ir_version, temp_dir):
import tensorflow as tf
from openvino.tools.ovc import convert_model
from openvino.runtime import compile_model
class LayerModel(tf.Module):
@@ -767,7 +798,6 @@ class TestMoConvertTF(CommonMOConvertTest):
def test_memory_loss(self, ie_device, precision, ir_version, temp_dir):
# This test checks that the memory allocated for constants
# is not lost after returning the model from convert_model() method.
import tensorflow as tf
tf.compat.v1.reset_default_graph()
from openvino.tools.ovc import convert_model
@@ -820,7 +850,6 @@ class TestMoConvertTF(CommonMOConvertTest):
assert CommonLayerTest().compare_ie_results_with_framework(ov_infer1, {"add:0": [2.6, 9.6, 12.4]}, eps)
def test_scalar(self, ie_device, precision, ir_version, temp_dir):
import tensorflow as tf
tf.compat.v1.reset_default_graph()
from openvino.tools.ovc import convert_model
@@ -861,7 +890,6 @@ class TestMoConvertTF(CommonMOConvertTest):
assert CommonLayerTest().compare_ie_results_with_framework(ov_infer, {"Identity:0": 3.2}, eps)
def test_unnamed_variable(self, ie_device, precision, ir_version, temp_dir):
import tensorflow as tf
tf.compat.v1.reset_default_graph()
from openvino.tools.ovc import convert_model
@@ -905,7 +933,6 @@ class TFConvertTest(unittest.TestCase):
@pytest.mark.nightly
@pytest.mark.precommit
def test_tf_function_no_signature(self):
import tensorflow as tf
from openvino.tools.ovc import convert_model
@tf.function()
@@ -920,8 +947,6 @@ class TFConvertTest(unittest.TestCase):
class TestTFLoadByModel(unittest.TestCase):
def test_load_by_model_tf_graph_iterator(self):
def simple_tf_model():
import tensorflow as tf
tf.compat.v1.reset_default_graph()
with tf.compat.v1.Session() as sess:
@@ -955,3 +980,37 @@ class TestTFConvertRaises(unittest.TestCase):
# check that it accepts specified names as is without parsing into 2 different inputs
with self.assertRaisesRegex(Exception, 'No node with name Input1\[1, 2, 3\],Input2\[1, 2, 3\]'):
convert_model(tf_model, input='Input1[1, 2, 3],Input2[1, 2, 3]')
class TestTFConversionParams(CommonMOConvertTest):
test_data = [
{'params_test': {'input': [tf.shape(tf.zeros((2, 3, 4))), tf.zeros((2, 3, 4)).shape, tf.TensorShape((2, 3, 4))]},
'fw_model': create_tf_model_three_inputs([1, 2, 3, 4]),
'ref_model': create_ref_model_three_inputs([2, 3, 4])},
{'params_test': {'input': [tf.float32, tf.float32, tf.float32]},
'fw_model': create_tf_model_three_inputs([2, 3], tf.int32),
'ref_model': create_ref_model_three_inputs([2, 3], np.float32)},
{'params_test': {'input': tf.shape(tf.zeros((5, 8, 2)))},
'fw_model': create_tf_model_single_input(),
'ref_model': create_ref_model_single_input([5, 8, 2])},
{'params_test': {'input': tf.zeros((9, 2)).shape},
'fw_model': create_tf_model_single_input(),
'ref_model': create_ref_model_single_input([9, 2])},
{'params_test': {'input': tf.TensorShape((4, 8, 3))},
'fw_model': create_tf_model_single_input(),
'ref_model': create_ref_model_single_input([4, 8, 3])},
{'params_test': {'input': tf.int32},
'fw_model': create_tf_model_single_input(),
'ref_model': create_ref_model_single_input([1, 2, 3, 4], np.int32)}
]
@pytest.mark.parametrize("params", test_data)
@pytest.mark.nightly
def test_mo_convert_tf_model(self, params, ie_device, precision, ir_version,
temp_dir, use_new_frontend, use_old_api):
fw_model = params['fw_model']
test_params = params['params_test']
ref_model = params['ref_model']
test_params.update({'input_model': fw_model})
self._test_by_ref_graph(temp_dir, test_params, ref_model, compare_tensor_names=False)