Daria Ilina
e4f44b19fd
* Mark all failed ONNX layer tests as XFail
* Add additional xfailed marks
* Add one more failed tests into XFail
* Add conditions for CPU/GPU failures
* Revert "Add conditions for CPU/GPU failures"
This reverts commit 790524c59c.
* Add failures separation for CPU/GPU
* Replace all xfail with skip
232 lines
10 KiB
Python
232 lines
10 KiB
Python
# Copyright (C) 2018-2023 Intel Corporation
|
|
# SPDX-License-Identifier: Apache-2.0
|
|
|
|
import numpy as np
|
|
import pytest
|
|
from common.onnx_layer_test_class import OnnxRuntimeLayerTest
|
|
|
|
|
|
class TestConvTranspose(OnnxRuntimeLayerTest):
|
|
def _prepare_input(self, inputs_dict):
|
|
for input in inputs_dict.keys():
|
|
inputs_dict[input] = np.random.randn(*inputs_dict[input]).astype(np.float32)
|
|
return inputs_dict
|
|
|
|
def create_conv_transpose(self, ir_version, input_shape, output_shape, kernel_shape, strides,
|
|
group=1,
|
|
dilations=None, pads=None, force_output_shape=False,
|
|
output_padding=None, bias=False,
|
|
auto_pad=None):
|
|
#
|
|
# Create ONNX model
|
|
#
|
|
|
|
import onnx
|
|
from onnx import helper
|
|
from onnx import TensorProto
|
|
|
|
input = helper.make_tensor_value_info('input', TensorProto.FLOAT, input_shape)
|
|
output = helper.make_tensor_value_info('output', TensorProto.FLOAT, output_shape)
|
|
|
|
weights = np.random.randn(*kernel_shape).astype(float)
|
|
|
|
node_weights_def = onnx.helper.make_node(
|
|
'Constant',
|
|
inputs=[],
|
|
outputs=['kernel'],
|
|
value=helper.make_tensor(
|
|
name='const_tensor',
|
|
data_type=TensorProto.FLOAT,
|
|
dims=weights.shape,
|
|
vals=weights.flatten(),
|
|
),
|
|
)
|
|
|
|
conv_attrs = {
|
|
'strides': strides,
|
|
'group': group,
|
|
'kernel_shape': kernel_shape[2:], # As we have NCHW layout
|
|
}
|
|
|
|
if pads is not None:
|
|
if not force_output_shape:
|
|
conv_attrs.update({'pads': pads})
|
|
else:
|
|
pads = np.zeros(2 * (len(input_shape) - 2))
|
|
_pads = np.array(pads).reshape([2, -1])
|
|
if output_padding is not None:
|
|
conv_attrs.update({'output_padding': output_padding})
|
|
if dilations is not None:
|
|
conv_attrs.update({'dilations': dilations})
|
|
else:
|
|
dilations = np.ones(len(input_shape) - 2)
|
|
if force_output_shape:
|
|
conv_attrs.update({'output_shape': output_shape[2:]})
|
|
|
|
if auto_pad:
|
|
conv_attrs.update({'auto_pad': auto_pad})
|
|
|
|
nodes = [node_weights_def]
|
|
if bias:
|
|
bias_const = np.random.randint(-10, 10, kernel_shape[0]).astype(np.float32)
|
|
|
|
node_bias_def = onnx.helper.make_node(
|
|
'Constant',
|
|
inputs=[],
|
|
outputs=['bias'],
|
|
value=helper.make_tensor(
|
|
name='const_tensor',
|
|
data_type=TensorProto.FLOAT,
|
|
dims=bias_const.shape,
|
|
vals=bias_const.flatten(),
|
|
),
|
|
)
|
|
node_conv_transpose = onnx.helper.make_node(
|
|
'ConvTranspose',
|
|
inputs=['input', 'kernel', 'bias'],
|
|
outputs=['output'],
|
|
**conv_attrs
|
|
)
|
|
nodes.extend([node_bias_def, node_conv_transpose])
|
|
else:
|
|
node_conv_transpose = onnx.helper.make_node(
|
|
'ConvTranspose',
|
|
inputs=['input', 'kernel'],
|
|
outputs=['output'],
|
|
**conv_attrs
|
|
)
|
|
nodes.append(node_conv_transpose)
|
|
|
|
# Create the graph (GraphProto)
|
|
graph_def = helper.make_graph(
|
|
nodes,
|
|
'test_conv_transpose_model',
|
|
[input],
|
|
[output],
|
|
)
|
|
|
|
# Create the model (ModelProto)
|
|
onnx_net = helper.make_model(graph_def, producer_name='test_conv_transpose_model')
|
|
|
|
#
|
|
# Create reference IR net
|
|
# Please, specify 'type': 'Input' for input node
|
|
# Moreover, do not forget to validate ALL layer attributes!!!
|
|
#
|
|
ref_net = None
|
|
|
|
return onnx_net, ref_net
|
|
|
|
common_tests_4D_precommit = [
|
|
pytest.param(dict(input_shape=[1, 3, 10, 10], output_shape=[1, 3, 12, 12],
|
|
kernel_shape=[3, 3, 2, 2], strides=[1, 1], dilations=[2, 2]),
|
|
marks=pytest.mark.skip(reason="Skipped until fixed")),
|
|
pytest.param(dict(input_shape=[1, 2, 20, 20], output_shape=[1, 2, 85, 85],
|
|
kernel_shape=[2, 1, 8, 8], strides=[4, 4], group=2,
|
|
output_padding=[1, 1]),
|
|
marks=pytest.mark.skip(reason="Skipped until fixed"))
|
|
]
|
|
|
|
common_tests_4D = [
|
|
dict(input_shape=[1, 3, 10, 10], output_shape=[1, 3, 11, 11],
|
|
kernel_shape=[3, 3, 2, 2], strides=[1, 1]),
|
|
dict(input_shape=[1, 3, 10, 10], output_shape=[1, 3, 20, 20],
|
|
kernel_shape=[3, 3, 2, 2], strides=[2, 2]),
|
|
dict(input_shape=[1, 3, 10, 10], output_shape=[1, 3, 12, 12],
|
|
kernel_shape=[3, 3, 2, 2], strides=[1, 1], dilations=[2, 2]),
|
|
dict(input_shape=[1, 3, 10, 10], output_shape=[1, 3, 21, 21],
|
|
kernel_shape=[3, 3, 2, 2], strides=[2, 2], dilations=[2, 2]),
|
|
dict(input_shape=[1, 2, 20, 20], output_shape=[1, 2, 85, 85],
|
|
kernel_shape=[2, 1, 8, 8], strides=[4, 4], group=2, output_padding=[1, 1]),
|
|
]
|
|
|
|
explicit_pads_tests_4D = common_tests_4D + [
|
|
dict(input_shape=[1, 2, 20, 20], output_shape=[1, 2, 80, 80],
|
|
kernel_shape=[2, 1, 8, 8], strides=[4, 4], group=2, pads=[2, 2, 2, 2]),
|
|
dict(input_shape=[1, 2, 20, 20], output_shape=[1, 2, 87, 87],
|
|
kernel_shape=[2, 1, 8, 8], strides=[4, 4], group=2, pads=[2, 2, 2, 2],
|
|
dilations=[2, 2]),
|
|
dict(input_shape=[1, 2, 20, 20], output_shape=[1, 2, 80, 80],
|
|
kernel_shape=[2, 1, 8, 8], strides=[4, 4], group=2, pads=[2, 2, 2, 2],
|
|
force_output_shape=True),
|
|
]
|
|
|
|
valid_auto_pad_tests_4D = common_tests_4D + [
|
|
dict(input_shape=[1, 2, 20, 20], output_shape=[1, 2, 84, 84],
|
|
kernel_shape=[2, 1, 8, 8], strides=[4, 4], group=2),
|
|
dict(input_shape=[1, 2, 20, 20], output_shape=[1, 2, 91, 91],
|
|
kernel_shape=[2, 1, 8, 8], strides=[4, 4], group=2, dilations=[2, 2]),
|
|
dict(input_shape=[1, 2, 20, 20], output_shape=[1, 2, 80, 80],
|
|
kernel_shape=[2, 1, 8, 8], strides=[4, 4], group=2, force_output_shape=True),
|
|
]
|
|
|
|
same_auto_pad_tests_4D = [
|
|
dict(input_shape=[1, 3, 10, 10], output_shape=[1, 3, 10, 10],
|
|
kernel_shape=[3, 3, 2, 2], strides=[1, 1]),
|
|
dict(input_shape=[1, 3, 10, 10], output_shape=[1, 3, 20, 20],
|
|
kernel_shape=[3, 3, 2, 2], strides=[2, 2]),
|
|
dict(input_shape=[1, 2, 20, 20], output_shape=[1, 2, 80, 80],
|
|
kernel_shape=[2, 1, 8, 8], strides=[4, 4], group=2),
|
|
dict(input_shape=[1, 3, 10, 10], output_shape=[1, 3, 10, 10],
|
|
kernel_shape=[3, 3, 2, 2], strides=[1, 1], dilations=[2, 2]),
|
|
dict(input_shape=[1, 3, 10, 10], output_shape=[1, 3, 20, 20],
|
|
kernel_shape=[3, 3, 2, 2], strides=[2, 2], dilations=[2, 2]),
|
|
dict(input_shape=[1, 2, 20, 20], output_shape=[1, 2, 80, 80],
|
|
kernel_shape=[2, 1, 8, 8], strides=[4, 4], group=2, dilations=[2, 2]),
|
|
dict(input_shape=[1, 2, 20, 20], output_shape=[1, 2, 80, 80],
|
|
kernel_shape=[2, 1, 8, 8], strides=[4, 4], group=2, force_output_shape=True),
|
|
dict(input_shape=[1, 2, 20, 20], output_shape=[1, 2, 81, 81],
|
|
kernel_shape=[2, 1, 8, 8], strides=[4, 4], group=2, output_padding=[1, 1]),
|
|
]
|
|
|
|
@pytest.mark.parametrize("params", common_tests_4D_precommit)
|
|
@pytest.mark.parametrize("bias", [False, True])
|
|
@pytest.mark.parametrize("auto_pad", ["NOTSET"])
|
|
@pytest.mark.precommit
|
|
def test_conv_transpose_4D_precommit(self, params, bias, ie_device, precision, ir_version,
|
|
auto_pad, temp_dir, use_old_api):
|
|
if ie_device == 'GPU' and 'dilations' in params:
|
|
pytest.xfail('dilations are not supported on GPU')
|
|
self._test(*self.create_conv_transpose(**params, ir_version=ir_version, bias=bias,
|
|
auto_pad=auto_pad),
|
|
ie_device, precision, ir_version, temp_dir=temp_dir, use_old_api=use_old_api)
|
|
|
|
@pytest.mark.parametrize("params", explicit_pads_tests_4D)
|
|
@pytest.mark.parametrize("bias", [False, True])
|
|
@pytest.mark.parametrize("auto_pad", ["NOTSET"])
|
|
@pytest.mark.nightly
|
|
@pytest.mark.skip(reason='GREEN_SUITE')
|
|
def test_conv_transpose_4D(self, params, bias, ie_device, precision, ir_version, auto_pad,
|
|
temp_dir, use_old_api):
|
|
if ie_device == 'GPU' and 'dilations' in params:
|
|
pytest.xfail('dilations are not supported on GPU')
|
|
self._test(*self.create_conv_transpose(**params, ir_version=ir_version, bias=bias,
|
|
auto_pad=auto_pad),
|
|
ie_device, precision, ir_version, temp_dir=temp_dir, use_old_api=use_old_api)
|
|
|
|
@pytest.mark.parametrize("params", valid_auto_pad_tests_4D)
|
|
@pytest.mark.parametrize("bias", [False, True])
|
|
@pytest.mark.parametrize("auto_pad", ["VALID"])
|
|
@pytest.mark.nightly
|
|
@pytest.mark.skip(reason='GREEN_SUITE')
|
|
def test_conv_transpose_valid_auto_pad_4D(self, params, bias, ie_device, precision, ir_version,
|
|
auto_pad, temp_dir, use_old_api):
|
|
if ie_device == 'GPU' and 'dilations' in params:
|
|
pytest.xfail('dilations are not supported on GPU')
|
|
self._test(*self.create_conv_transpose(**params, ir_version=ir_version, bias=bias,
|
|
auto_pad=auto_pad),
|
|
ie_device, precision, ir_version, temp_dir=temp_dir, use_old_api=use_old_api)
|
|
|
|
@pytest.mark.parametrize("params", same_auto_pad_tests_4D)
|
|
@pytest.mark.parametrize("bias", [False, True])
|
|
@pytest.mark.parametrize("auto_pad", ["SAME_UPPER", "SAME_LOWER"])
|
|
@pytest.mark.nightly
|
|
@pytest.mark.skip(reason='GREEN_SUITE')
|
|
def test_conv_transpose_same_auto_pad_4D(self, params, bias, ie_device, precision, ir_version,
|
|
auto_pad, temp_dir, use_old_api):
|
|
if ie_device == 'GPU' and 'dilations' in params:
|
|
pytest.xfail('dilations are not supported on GPU')
|
|
self._test(*self.create_conv_transpose(**params, ir_version=ir_version, bias=bias,
|
|
auto_pad=auto_pad),
|
|
ie_device, precision, ir_version, temp_dir=temp_dir, use_old_api=use_old_api)
|