Skip to content

P
Paddle

PaddlePaddle / Paddle
大约 2 年 前同步成功

通知 2325
Star 20933
Fork 5424
P
Paddle
未验证 提交 d26ae9ad 编写于 作者: L LielinJiang 提交者： GitHub
浏览文件
操作

Update conv_transpose api (#26427) 

* update conv_transpose api
上级 c037d625
显示空白变更内容
内联 并排
Showing with 471 addition and 505 deletion
paddle/fluid/operators/conv_transpose_op.cc
浏览文件 @ d26ae9ad
...@@ -37,6 +37,8 @@ void ConvTransposeOp::InferShape(framework::InferShapeContext* ctx) const { ...@@ -37,6 +37,8 @@ void ConvTransposeOp::InferShape(framework::InferShapeContext* ctx) const {
auto filter_dims = ctx->GetInputDim("Filter"); auto filter_dims = ctx->GetInputDim("Filter");
std::vector<int> output_size = std::vector<int> output_size =
ctx->Attrs().Get<std::vector<int>>("output_size"); ctx->Attrs().Get<std::vector<int>>("output_size");
std::vector<int> output_padding =
ctx->Attrs().Get<std::vector<int>>("output_padding");
std::vector<int> strides = ctx->Attrs().Get<std::vector<int>>("strides"); std::vector<int> strides = ctx->Attrs().Get<std::vector<int>>("strides");
std::vector<int> paddings = ctx->Attrs().Get<std::vector<int>>("paddings"); std::vector<int> paddings = ctx->Attrs().Get<std::vector<int>>("paddings");
std::vector<int> dilations = ctx->Attrs().Get<std::vector<int>>("dilations"); std::vector<int> dilations = ctx->Attrs().Get<std::vector<int>>("dilations");
...@@ -78,6 +80,12 @@ void ConvTransposeOp::InferShape(framework::InferShapeContext* ctx) const { ...@@ -78,6 +80,12 @@ void ConvTransposeOp::InferShape(framework::InferShapeContext* ctx) const {
platform::errors::InvalidArgument( platform::errors::InvalidArgument(
"The Attr(output_size) and Attr(stride) of Op(conv_transpose) " "The Attr(output_size) and Attr(stride) of Op(conv_transpose) "
"should be the same.")); "should be the same."));
if (output_padding.size())
PADDLE_ENFORCE_EQ(
output_padding.size(), strides.size(),
platform::errors::InvalidArgument(
"The Attr(output_padding) and Attr(stride) of Op(conv_transpose) "
"should be the same."));
const int64_t C = const int64_t C =
(data_layout != DataLayout::kNHWC ? in_dims[1] (data_layout != DataLayout::kNHWC ? in_dims[1]
...@@ -136,6 +144,27 @@ void ConvTransposeOp::InferShape(framework::InferShapeContext* ctx) const { ...@@ -136,6 +144,27 @@ void ConvTransposeOp::InferShape(framework::InferShapeContext* ctx) const {
infer_shape + strides[i])); infer_shape + strides[i]));
} }
output_shape.push_back(output_size[i]); output_shape.push_back(output_size[i]);
} else if (output_padding.size()) {
if (ctx->IsRuntime()) {
PADDLE_ENFORCE_GE(
output_padding[i], 0,
platform::errors::InvalidArgument(
"output_padding of Op(ConvTransposeOp) should not be "
"less than the 0. But received output_padding = "
"[%s], whose dim %d is less than 0",
framework::make_ddim(output_padding), i));
PADDLE_ENFORCE_LT(
output_padding[i], std::max(strides[i], dilations[i]),
platform::errors::InvalidArgument(
"output_padding of Op(ConvTransposeOp) should be less "
"than either stride or dilation. But received output_size = "
"[%s], "
"whose dim %d is not less than either stride (%d) or "
"dilation (%d)",
framework::make_ddim(output_size), i, strides[i],
dilations[i]));
}
output_shape.push_back((infer_shape + output_padding[i]));
} else { } else {
output_shape.push_back(infer_shape); output_shape.push_back(infer_shape);
} }
...@@ -223,10 +252,14 @@ void Conv2DTransposeOpMaker::Make() { ...@@ -223,10 +252,14 @@ void Conv2DTransposeOpMaker::Make() {
"The format of output tensor is X (one-dimensional) of size equal" "The format of output tensor is X (one-dimensional) of size equal"
"to the number of output channels. Only used with MKL-DNN.") "to the number of output channels. Only used with MKL-DNN.")
.AsDispensable(); .AsDispensable();
AddOutput("Output", AddOutput("Output",
"(Tensor) The output tensor of convolution transpose operator. " "(Tensor) The output tensor of convolution transpose operator. "
"The format of output tensor is the same as input tensor."); "The format of output tensor is the same as input tensor.");
AddAttr<std::vector<int>>("output_padding",
"(vector<int> default: []), Additional size added "
"to one side of each dimension in the output "
"shape")
.SetDefault({});
AddAttr<std::vector<int>>("output_size", AddAttr<std::vector<int>>("output_size",
"(vector<int> default: []), the " "(vector<int> default: []), the "
"size of the output tensor") "size of the output tensor")
...@@ -338,6 +371,11 @@ void Conv3DTransposeOpMaker::Make() { ...@@ -338,6 +371,11 @@ void Conv3DTransposeOpMaker::Make() {
"Where N is batch size, C is " "Where N is batch size, C is "
"the number of channels, D is the depth of the feature, H is the " "the number of channels, D is the depth of the feature, H is the "
"height of the feature, and W is the width of the feature."); "height of the feature, and W is the width of the feature.");
AddAttr<std::vector<int>>("output_padding",
"(vector<int> default: []), Additional size added "
"to one side of each dimension in the output "
"shape")
.SetDefault({});
AddAttr<std::vector<int>>("output_size", AddAttr<std::vector<int>>("output_size",
"(vector<int> default: []), the " "(vector<int> default: []), the "
"size of the output tensor") "size of the output tensor")
......
python/paddle/fluid/tests/unittests/test_conv2d_transpose_layer.py
浏览文件 @ d26ae9ad
...@@ -29,13 +29,12 @@ class Conv2DTransposeTestCase(unittest.TestCase): ...@@ -29,13 +29,12 @@ class Conv2DTransposeTestCase(unittest.TestCase):
num_filters=8, num_filters=8,
filter_size=3, filter_size=3,
output_size=None, output_size=None,
output_padding=0,
padding=0, padding=0,
stride=1, stride=1,
dilation=1, dilation=1,
groups=1, groups=1,
act=None,
no_bias=False, no_bias=False,
use_cudnn=True,
data_format="NCHW", data_format="NCHW",
dtype="float32"): dtype="float32"):
super(Conv2DTransposeTestCase, self).__init__(methodName) super(Conv2DTransposeTestCase, self).__init__(methodName)
...@@ -45,14 +44,13 @@ class Conv2DTransposeTestCase(unittest.TestCase): ...@@ -45,14 +44,13 @@ class Conv2DTransposeTestCase(unittest.TestCase):
self.spartial_shape = spartial_shape self.spartial_shape = spartial_shape
self.filter_size = filter_size self.filter_size = filter_size
self.output_size = output_size self.output_size = output_size
self.output_padding = output_padding
self.padding = padding self.padding = padding
self.stride = stride self.stride = stride
self.dilation = dilation self.dilation = dilation
self.groups = groups self.groups = groups
self.act = act
self.no_bias = no_bias self.no_bias = no_bias
self.use_cudnn = use_cudnn
self.data_format = data_format self.data_format = data_format
self.dtype = dtype self.dtype = dtype
...@@ -93,6 +91,7 @@ class Conv2DTransposeTestCase(unittest.TestCase): ...@@ -93,6 +91,7 @@ class Conv2DTransposeTestCase(unittest.TestCase):
bias_attr = False bias_attr = False
else: else:
bias_attr = I.NumpyArrayInitializer(self.bias) bias_attr = I.NumpyArrayInitializer(self.bias)
y_var = fluid.layers.conv2d_transpose( y_var = fluid.layers.conv2d_transpose(
x_var, x_var,
self.num_filters, self.num_filters,
...@@ -104,8 +103,6 @@ class Conv2DTransposeTestCase(unittest.TestCase): ...@@ -104,8 +103,6 @@ class Conv2DTransposeTestCase(unittest.TestCase):
groups=self.groups, groups=self.groups,
param_attr=weight_attr, param_attr=weight_attr,
bias_attr=bias_attr, bias_attr=bias_attr,
use_cudnn=self.use_cudnn,
act=self.act,
data_format=self.data_format) data_format=self.data_format)
feed_dict = {"input": self.input} feed_dict = {"input": self.input}
exe = fluid.Executor(place) exe = fluid.Executor(place)
...@@ -125,17 +122,22 @@ class Conv2DTransposeTestCase(unittest.TestCase): ...@@ -125,17 +122,22 @@ class Conv2DTransposeTestCase(unittest.TestCase):
"weight", self.weight_shape, dtype=self.dtype) "weight", self.weight_shape, dtype=self.dtype)
b_var = fluid.data( b_var = fluid.data(
"bias", (self.num_filters, ), dtype=self.dtype) "bias", (self.num_filters, ), dtype=self.dtype)
y_var = F.conv2d_transpose(
if self.output_padding != 0:
output_size = None
else:
output_size = self.output_size
y_var = F.conv_transpose2d(
x_var, x_var,
w_var, w_var,
None if self.no_bias else b_var, None if self.no_bias else b_var,
output_size=self.output_size, output_size=output_size,
padding=self.padding, padding=self.padding,
output_padding=self.output_padding,
stride=self.stride, stride=self.stride,
dilation=self.dilation, dilation=self.dilation,
groups=self.groups, groups=self.groups,
act=self.act,
use_cudnn=self.use_cudnn,
data_format=self.data_format) data_format=self.data_format)
feed_dict = {"input": self.input, "weight": self.weight} feed_dict = {"input": self.input, "weight": self.weight}
if self.bias is not None: if self.bias is not None:
...@@ -147,32 +149,38 @@ class Conv2DTransposeTestCase(unittest.TestCase): ...@@ -147,32 +149,38 @@ class Conv2DTransposeTestCase(unittest.TestCase):
def paddle_nn_layer(self): def paddle_nn_layer(self):
x_var = dg.to_variable(self.input) x_var = dg.to_variable(self.input)
conv = nn.Conv2DTranspose(
if self.output_padding != 0:
output_size = None
else:
output_size = self.output_size
conv = nn.ConvTranspose2d(
self.num_channels, self.num_channels,
self.num_filters, self.num_filters,
self.filter_size, self.filter_size,
output_size=self.output_size,
padding=self.padding, padding=self.padding,
output_padding=self.output_padding,
stride=self.stride, stride=self.stride,
dilation=self.dilation, dilation=self.dilation,
groups=self.groups, groups=self.groups,
act=self.act, data_format=self.data_format)
use_cudnn=self.use_cudnn,
data_format=self.data_format,
dtype=self.dtype)
conv.weight.set_value(self.weight) conv.weight.set_value(self.weight)
if not self.no_bias: if not self.no_bias:
conv.bias.set_value(self.bias) conv.bias.set_value(self.bias)
y_var = conv(x_var) y_var = conv(x_var, output_size)
y_np = y_var.numpy() y_np = y_var.numpy()
return y_np return y_np
def _test_equivalence(self, place): def _test_equivalence(self, place):
place = fluid.CPUPlace() place = fluid.CPUPlace()
result1 = self.fluid_layer(place) result1 = self.fluid_layer(place)
result2 = self.functional(place) result2 = self.functional(place)
with dg.guard(place): with dg.guard(place):
result3 = self.paddle_nn_layer() result3 = self.paddle_nn_layer()
np.testing.assert_array_almost_equal(result1, result2) np.testing.assert_array_almost_equal(result1, result2)
np.testing.assert_array_almost_equal(result2, result3) np.testing.assert_array_almost_equal(result2, result3)
...@@ -194,7 +202,7 @@ class Conv2DTransposeErrorTestCase(Conv2DTransposeTestCase): ...@@ -194,7 +202,7 @@ class Conv2DTransposeErrorTestCase(Conv2DTransposeTestCase):
def add_cases(suite): def add_cases(suite):
suite.addTest(Conv2DTransposeTestCase(methodName='runTest', act="relu")) suite.addTest(Conv2DTransposeTestCase(methodName='runTest'))
suite.addTest( suite.addTest(
Conv2DTransposeTestCase( Conv2DTransposeTestCase(
methodName='runTest', stride=[1, 2], no_bias=True, dilation=2)) methodName='runTest', stride=[1, 2], no_bias=True, dilation=2))
...@@ -211,9 +219,6 @@ def add_cases(suite): ...@@ -211,9 +219,6 @@ def add_cases(suite):
suite.addTest( suite.addTest(
Conv2DTransposeTestCase( Conv2DTransposeTestCase(
methodName='runTest', padding="valid")) methodName='runTest', padding="valid"))
suite.addTest(
Conv2DTransposeTestCase(
methodName='runTest', padding='valid'))
suite.addTest( suite.addTest(
Conv2DTransposeTestCase( Conv2DTransposeTestCase(
methodName='runTest', filter_size=1, padding=(2, 3))) methodName='runTest', filter_size=1, padding=(2, 3)))
...@@ -240,15 +245,22 @@ def add_cases(suite): ...@@ -240,15 +245,22 @@ def add_cases(suite):
num_filters=6, num_filters=6,
num_channels=3, num_channels=3,
groups=3, groups=3,
use_cudnn=False,
act="sigmoid",
padding="valid")) padding="valid"))
suite.addTest(
Conv2DTransposeTestCase(
methodName='runTest',
num_filters=6,
num_channels=3,
spartial_shape=(7, 7),
filter_size=[5, 5],
groups=1,
padding=2,
stride=2,
output_size=[14, 14],
output_padding=[1, 1], ))
def add_error_cases(suite): def add_error_cases(suite):
suite.addTest(
Conv2DTransposeErrorTestCase(
methodName='runTest', use_cudnn="not_valid"))
suite.addTest( suite.addTest(
Conv2DTransposeErrorTestCase( Conv2DTransposeErrorTestCase(
methodName='runTest', num_channels=5, groups=2)) methodName='runTest', num_channels=5, groups=2))
......
python/paddle/fluid/tests/unittests/test_conv2d_transpose_op.py
浏览文件 @ d26ae9ad
...@@ -77,8 +77,13 @@ def conv2dtranspose_forward_naive(input_, filter_, attrs): ...@@ -77,8 +77,13 @@ def conv2dtranspose_forward_naive(input_, filter_, attrs):
output_size = attrs['output_size'] output_size = attrs['output_size']
out_h = output_size[0] + pad_h_0 + pad_h_1 out_h = output_size[0] + pad_h_0 + pad_h_1
out_w = output_size[1] + pad_w_0 + pad_w_1 out_w = output_size[1] + pad_w_0 + pad_w_1
out_pad_h = 0
out = np.zeros((in_n, out_c, out_h, out_w), dtype=input_.dtype) out_pad_w = 0
if 'output_padding' in attrs:
out_pad_h = attrs['output_padding'][0]
out_pad_w = attrs['output_padding'][1]
out = np.zeros(
(in_n, out_c, out_h + out_pad_h, out_w + out_pad_w), dtype=input_.dtype)
for n in range(in_n): for n in range(in_n):
for i in range(in_h): for i in range(in_h):
...@@ -99,7 +104,8 @@ def conv2dtranspose_forward_naive(input_, filter_, attrs): ...@@ -99,7 +104,8 @@ def conv2dtranspose_forward_naive(input_, filter_, attrs):
out[n, g * f_out_c + k, i1:i2:dilations[0], j1:j2: out[n, g * f_out_c + k, i1:i2:dilations[0], j1:j2:
dilations[1]] += tmp_out dilations[1]] += tmp_out
out = out[:, :, pad_h_0:out_h - pad_h_1, pad_w_0:out_w - pad_w_1] out = out[:, :, pad_h_0:out_h - pad_h_1 + out_pad_h, pad_w_0:out_w - pad_w_1
+ out_pad_w]
if attrs['data_format'] == 'NHWC': if attrs['data_format'] == 'NHWC':
out = np.transpose(out, [0, 2, 3, 1]) out = np.transpose(out, [0, 2, 3, 1])
return out return out
...@@ -114,6 +120,7 @@ class TestConv2dTransposeOp(OpTest): ...@@ -114,6 +120,7 @@ class TestConv2dTransposeOp(OpTest):
self.use_cudnn = False self.use_cudnn = False
self.use_mkldnn = False self.use_mkldnn = False
self.output_size = None self.output_size = None
self.output_padding = []
self.data_format = "NCHW" self.data_format = "NCHW"
self.pad = [0, 0] self.pad = [0, 0]
self.padding_algorithm = "EXPLICIT" self.padding_algorithm = "EXPLICIT"
...@@ -138,6 +145,9 @@ class TestConv2dTransposeOp(OpTest): ...@@ -138,6 +145,9 @@ class TestConv2dTransposeOp(OpTest):
if self.output_size is not None: if self.output_size is not None:
self.attrs['output_size'] = self.output_size self.attrs['output_size'] = self.output_size
if len(self.output_padding) > 0:
self.attrs['output_padding'] = self.output_padding
output = conv2dtranspose_forward_naive(input_, filter_, output = conv2dtranspose_forward_naive(input_, filter_,
self.attrs).astype(self.dtype) self.attrs).astype(self.dtype)
...@@ -290,6 +300,18 @@ class TestWithEvenUpsample(TestConv2dTransposeOp): ...@@ -290,6 +300,18 @@ class TestWithEvenUpsample(TestConv2dTransposeOp):
self.filter_size = [f_c, 6, 5, 5] self.filter_size = [f_c, 6, 5, 5]
class TestWithEvenUpsampleOutputPadding(TestConv2dTransposeOp):
def init_test_case(self):
self.pad = [2, 2]
self.stride = [2, 2]
self.groups = 1
self.dilations = [1, 1]
self.output_padding = [1, 1]
self.input_size = [2, 3, 7, 7] # NCHW
f_c = self.input_size[1]
self.filter_size = [f_c, 6, 5, 5]
class Test_NHWC(TestConv2dTransposeOp): class Test_NHWC(TestConv2dTransposeOp):
def init_test_case(self): def init_test_case(self):
self.pad = [0, 0] self.pad = [0, 0]
...@@ -375,6 +397,19 @@ class TestWithEvenUpsample_NHWC(TestConv2dTransposeOp): ...@@ -375,6 +397,19 @@ class TestWithEvenUpsample_NHWC(TestConv2dTransposeOp):
self.data_format = 'NHWC' self.data_format = 'NHWC'
class TestWithEvenUpsample_NHWC_output_padding(TestConv2dTransposeOp):
def init_test_case(self):
self.pad = [2, 2]
self.stride = [2, 2]
self.groups = 1
self.dilations = [1, 1]
self.output_padding = [1, 1]
self.input_size = [2, 7, 7, 3] # NHWC
f_c = self.input_size[-1]
self.filter_size = [f_c, 6, 5, 5]
self.data_format = 'NHWC'
# ------------ test_cudnn ------------ # ------------ test_cudnn ------------
@unittest.skipIf(not core.is_compiled_with_cuda(), @unittest.skipIf(not core.is_compiled_with_cuda(),
"core is not compiled with CUDA") "core is not compiled with CUDA")
......
python/paddle/fluid/tests/unittests/test_conv3d_transpose_layer.py
浏览文件 @ d26ae9ad
...@@ -33,9 +33,7 @@ class Conv3DTransposeTestCase(unittest.TestCase): ...@@ -33,9 +33,7 @@ class Conv3DTransposeTestCase(unittest.TestCase):
stride=1, stride=1,
dilation=1, dilation=1,
groups=1, groups=1,
act=None,
no_bias=False, no_bias=False,
use_cudnn=True,
data_format="NCDHW", data_format="NCDHW",
dtype="float32"): dtype="float32"):
super(Conv3DTransposeTestCase, self).__init__(methodName) super(Conv3DTransposeTestCase, self).__init__(methodName)
...@@ -50,9 +48,7 @@ class Conv3DTransposeTestCase(unittest.TestCase): ...@@ -50,9 +48,7 @@ class Conv3DTransposeTestCase(unittest.TestCase):
self.stride = stride self.stride = stride
self.dilation = dilation self.dilation = dilation
self.groups = groups self.groups = groups
self.act = act
self.no_bias = no_bias self.no_bias = no_bias
self.use_cudnn = use_cudnn
self.data_format = data_format self.data_format = data_format
self.dtype = dtype self.dtype = dtype
...@@ -104,8 +100,6 @@ class Conv3DTransposeTestCase(unittest.TestCase): ...@@ -104,8 +100,6 @@ class Conv3DTransposeTestCase(unittest.TestCase):
groups=self.groups, groups=self.groups,
param_attr=weight_attr, param_attr=weight_attr,
bias_attr=bias_attr, bias_attr=bias_attr,
use_cudnn=self.use_cudnn,
act=self.act,
data_format=self.data_format) data_format=self.data_format)
feed_dict = {"input": self.input} feed_dict = {"input": self.input}
exe = fluid.Executor(place) exe = fluid.Executor(place)
...@@ -125,7 +119,7 @@ class Conv3DTransposeTestCase(unittest.TestCase): ...@@ -125,7 +119,7 @@ class Conv3DTransposeTestCase(unittest.TestCase):
"weight", self.weight_shape, dtype=self.dtype) "weight", self.weight_shape, dtype=self.dtype)
b_var = fluid.data( b_var = fluid.data(
"bias", (self.num_filters, ), dtype=self.dtype) "bias", (self.num_filters, ), dtype=self.dtype)
y_var = F.conv3d_transpose( y_var = F.conv_transpose3d(
x_var, x_var,
w_var, w_var,
None if self.no_bias else b_var, None if self.no_bias else b_var,
...@@ -134,8 +128,6 @@ class Conv3DTransposeTestCase(unittest.TestCase): ...@@ -134,8 +128,6 @@ class Conv3DTransposeTestCase(unittest.TestCase):
stride=self.stride, stride=self.stride,
dilation=self.dilation, dilation=self.dilation,
groups=self.groups, groups=self.groups,
act=self.act,
use_cudnn=self.use_cudnn,
data_format=self.data_format) data_format=self.data_format)
feed_dict = {"input": self.input, "weight": self.weight} feed_dict = {"input": self.input, "weight": self.weight}
if self.bias is not None: if self.bias is not None:
...@@ -147,23 +139,19 @@ class Conv3DTransposeTestCase(unittest.TestCase): ...@@ -147,23 +139,19 @@ class Conv3DTransposeTestCase(unittest.TestCase):
def paddle_nn_layer(self): def paddle_nn_layer(self):
x_var = dg.to_variable(self.input) x_var = dg.to_variable(self.input)
conv = nn.Conv3DTranspose( conv = nn.ConvTranspose3d(
self.num_channels, self.num_channels,
self.num_filters, self.num_filters,
self.filter_size, self.filter_size,
output_size=self.output_size,
padding=self.padding, padding=self.padding,
stride=self.stride, stride=self.stride,
dilation=self.dilation, dilation=self.dilation,
groups=self.groups, groups=self.groups,
act=self.act, data_format=self.data_format)
use_cudnn=self.use_cudnn,
data_format=self.data_format,
dtype=self.dtype)
conv.weight.set_value(self.weight) conv.weight.set_value(self.weight)
if not self.no_bias: if not self.no_bias:
conv.bias.set_value(self.bias) conv.bias.set_value(self.bias)
y_var = conv(x_var) y_var = conv(x_var, self.output_size)
y_np = y_var.numpy() y_np = y_var.numpy()
return y_np return y_np
...@@ -194,7 +182,7 @@ class Conv3DTransposeErrorTestCase(Conv3DTransposeTestCase): ...@@ -194,7 +182,7 @@ class Conv3DTransposeErrorTestCase(Conv3DTransposeTestCase):
def add_cases(suite): def add_cases(suite):
suite.addTest(Conv3DTransposeTestCase(methodName='runTest', act="tanh")) suite.addTest(Conv3DTransposeTestCase(methodName='runTest'))
suite.addTest( suite.addTest(
Conv3DTransposeTestCase( Conv3DTransposeTestCase(
methodName='runTest', stride=[1, 2, 1], dilation=2, no_bias=True)) methodName='runTest', stride=[1, 2, 1], dilation=2, no_bias=True))
...@@ -240,15 +228,10 @@ def add_cases(suite): ...@@ -240,15 +228,10 @@ def add_cases(suite):
num_filters=6, num_filters=6,
num_channels=3, num_channels=3,
groups=3, groups=3,
use_cudnn=False,
act="sigmoid",
padding="valid")) padding="valid"))
def add_error_cases(suite): def add_error_cases(suite):
suite.addTest(
Conv3DTransposeErrorTestCase(
methodName='runTest', use_cudnn="not_valid"))
suite.addTest( suite.addTest(
Conv3DTransposeErrorTestCase( Conv3DTransposeErrorTestCase(
methodName='runTest', num_channels=5, groups=2)) methodName='runTest', num_channels=5, groups=2))
......
python/paddle/fluid/tests/unittests/test_functional_conv2d_transpose.py
浏览文件 @ d26ae9ad
...@@ -37,8 +37,6 @@ class TestFunctionalConv2D(TestCase): ...@@ -37,8 +37,6 @@ class TestFunctionalConv2D(TestCase):
self.dilation = 1 self.dilation = 1
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NHWC" self.data_format = "NHWC"
def prepare(self): def prepare(self):
...@@ -90,8 +88,6 @@ class TestFunctionalConv2D(TestCase): ...@@ -90,8 +88,6 @@ class TestFunctionalConv2D(TestCase):
param_attr=I.NumpyArrayInitializer(self.weight), param_attr=I.NumpyArrayInitializer(self.weight),
bias_attr=False bias_attr=False
if self.no_bias else I.NumpyArrayInitializer(self.bias), if self.no_bias else I.NumpyArrayInitializer(self.bias),
use_cudnn=self.use_cudnn,
act=self.act,
data_format=self.data_format) data_format=self.data_format)
exe = fluid.Executor(self.place) exe = fluid.Executor(self.place)
exe.run(start) exe.run(start)
...@@ -115,7 +111,7 @@ class TestFunctionalConv2D(TestCase): ...@@ -115,7 +111,7 @@ class TestFunctionalConv2D(TestCase):
"weight", self.weight.shape, dtype=self.dtype) "weight", self.weight.shape, dtype=self.dtype)
if not self.no_bias: if not self.no_bias:
bias = fluid.data("bias", self.bias.shape, dtype=self.dtype) bias = fluid.data("bias", self.bias.shape, dtype=self.dtype)
y = F.conv2d_transpose( y = F.conv_transpose2d(
x, x,
weight, weight,
None if self.no_bias else bias, None if self.no_bias else bias,
...@@ -124,9 +120,7 @@ class TestFunctionalConv2D(TestCase): ...@@ -124,9 +120,7 @@ class TestFunctionalConv2D(TestCase):
stride=self.stride, stride=self.stride,
dilation=self.dilation, dilation=self.dilation,
groups=self.groups, groups=self.groups,
act=self.act, data_format=self.data_format)
data_format=self.data_format,
use_cudnn=self.use_cudnn)
exe = fluid.Executor(self.place) exe = fluid.Executor(self.place)
exe.run(start) exe.run(start)
feed_dict = {"input": self.input, "weight": self.weight} feed_dict = {"input": self.input, "weight": self.weight}
...@@ -140,7 +134,7 @@ class TestFunctionalConv2D(TestCase): ...@@ -140,7 +134,7 @@ class TestFunctionalConv2D(TestCase):
x = dg.to_variable(self.input) x = dg.to_variable(self.input)
weight = dg.to_variable(self.weight) weight = dg.to_variable(self.weight)
bias = None if self.no_bias else dg.to_variable(self.bias) bias = None if self.no_bias else dg.to_variable(self.bias)
y = F.conv2d_transpose( y = F.conv_transpose2d(
x, x,
weight, weight,
bias, bias,
...@@ -148,10 +142,8 @@ class TestFunctionalConv2D(TestCase): ...@@ -148,10 +142,8 @@ class TestFunctionalConv2D(TestCase):
padding=self.padding, padding=self.padding,
stride=self.stride, stride=self.stride,
dilation=self.dilation, dilation=self.dilation,
act=self.act,
groups=self.groups, groups=self.groups,
data_format=self.data_format, data_format=self.data_format)
use_cudnn=self.use_cudnn)
out = y.numpy() out = y.numpy()
return out return out
...@@ -189,8 +181,6 @@ class TestFunctionalConv2DError(TestCase): ...@@ -189,8 +181,6 @@ class TestFunctionalConv2DError(TestCase):
self.dilation = 1 self.dilation = 1
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NHWC" self.data_format = "NHWC"
def test_exception(self): def test_exception(self):
...@@ -225,7 +215,7 @@ class TestFunctionalConv2DError(TestCase): ...@@ -225,7 +215,7 @@ class TestFunctionalConv2DError(TestCase):
"weight", self.weight_shape, dtype=self.dtype) "weight", self.weight_shape, dtype=self.dtype)
if not self.no_bias: if not self.no_bias:
bias = fluid.data("bias", self.bias_shape, dtype=self.dtype) bias = fluid.data("bias", self.bias_shape, dtype=self.dtype)
y = F.conv2d_transpose( y = F.conv_transpose2d(
x, x,
weight, weight,
None if self.no_bias else bias, None if self.no_bias else bias,
...@@ -234,9 +224,7 @@ class TestFunctionalConv2DError(TestCase): ...@@ -234,9 +224,7 @@ class TestFunctionalConv2DError(TestCase):
stride=self.stride, stride=self.stride,
dilation=self.dilation, dilation=self.dilation,
groups=self.groups, groups=self.groups,
act=self.act, data_format=self.data_format)
data_format=self.data_format,
use_cudnn=self.use_cudnn)
class TestFunctionalConv2DCase2(TestFunctionalConv2D): class TestFunctionalConv2DCase2(TestFunctionalConv2D):
...@@ -249,8 +237,6 @@ class TestFunctionalConv2DCase2(TestFunctionalConv2D): ...@@ -249,8 +237,6 @@ class TestFunctionalConv2DCase2(TestFunctionalConv2D):
self.dilation = 1 self.dilation = 1
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NHWC" self.data_format = "NHWC"
...@@ -264,8 +250,6 @@ class TestFunctionalConv2DCase3(TestFunctionalConv2D): ...@@ -264,8 +250,6 @@ class TestFunctionalConv2DCase3(TestFunctionalConv2D):
self.dilation = 1 self.dilation = 1
self.groups = 1 self.groups = 1
self.no_bias = True self.no_bias = True
self.act = None
self.use_cudnn = True
self.data_format = "NCHW" self.data_format = "NCHW"
...@@ -279,8 +263,6 @@ class TestFunctionalConv2DCase4(TestFunctionalConv2D): ...@@ -279,8 +263,6 @@ class TestFunctionalConv2DCase4(TestFunctionalConv2D):
self.dilation = 1 self.dilation = 1
self.groups = 2 self.groups = 2
self.no_bias = False self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NHWC" self.data_format = "NHWC"
...@@ -294,8 +276,6 @@ class TestFunctionalConv2DCase5(TestFunctionalConv2D): ...@@ -294,8 +276,6 @@ class TestFunctionalConv2DCase5(TestFunctionalConv2D):
self.dilation = 1 self.dilation = 1
self.groups = 2 self.groups = 2
self.no_bias = False self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NHWC" self.data_format = "NHWC"
...@@ -309,8 +289,6 @@ class TestFunctionalConv2DCase6(TestFunctionalConv2D): ...@@ -309,8 +289,6 @@ class TestFunctionalConv2DCase6(TestFunctionalConv2D):
self.dilation = (2, 1) self.dilation = (2, 1)
self.groups = 2 self.groups = 2
self.no_bias = False self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NHWC" self.data_format = "NHWC"
...@@ -324,8 +302,6 @@ class TestFunctionalConv2DCase7(TestFunctionalConv2D): ...@@ -324,8 +302,6 @@ class TestFunctionalConv2DCase7(TestFunctionalConv2D):
self.dilation = 1 self.dilation = 1
self.groups = 4 self.groups = 4
self.no_bias = False self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = False
self.data_format = "NHWC" self.data_format = "NHWC"
...@@ -340,8 +316,6 @@ class TestFunctionalConv2DCase8(TestFunctionalConv2D): ...@@ -340,8 +316,6 @@ class TestFunctionalConv2DCase8(TestFunctionalConv2D):
self.dilation = 1 self.dilation = 1
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NCHW" self.data_format = "NCHW"
...@@ -355,8 +329,6 @@ class TestFunctionalConv2DCase9(TestFunctionalConv2D): ...@@ -355,8 +329,6 @@ class TestFunctionalConv2DCase9(TestFunctionalConv2D):
self.dilation = 1 self.dilation = 1
self.groups = 2 self.groups = 2
self.no_bias = False self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NHWC" self.data_format = "NHWC"
...@@ -370,8 +342,6 @@ class TestFunctionalConv2DCase10(TestFunctionalConv2D): ...@@ -370,8 +342,6 @@ class TestFunctionalConv2DCase10(TestFunctionalConv2D):
self.dilation = 1 self.dilation = 1
self.groups = 2 self.groups = 2
self.no_bias = False self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NCHW" self.data_format = "NCHW"
...@@ -385,8 +355,6 @@ class TestFunctionalConv2DCase11(TestFunctionalConv2D): ...@@ -385,8 +355,6 @@ class TestFunctionalConv2DCase11(TestFunctionalConv2D):
self.dilation = 1 self.dilation = 1
self.groups = 2 self.groups = 2
self.no_bias = False self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NCHW" self.data_format = "NCHW"
...@@ -400,8 +368,6 @@ class TestFunctionalConv2DCase12(TestFunctionalConv2D): ...@@ -400,8 +368,6 @@ class TestFunctionalConv2DCase12(TestFunctionalConv2D):
self.dilation = 1 self.dilation = 1
self.groups = 2 self.groups = 2
self.no_bias = False self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NCHW" self.data_format = "NCHW"
...@@ -415,8 +381,6 @@ class TestFunctionalConv2DErrorCase2(TestFunctionalConv2DError): ...@@ -415,8 +381,6 @@ class TestFunctionalConv2DErrorCase2(TestFunctionalConv2DError):
self.dilation = 1 self.dilation = 1
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NHWC" self.data_format = "NHWC"
...@@ -430,8 +394,6 @@ class TestFunctionalConv2DErrorCase3(TestFunctionalConv2DError): ...@@ -430,8 +394,6 @@ class TestFunctionalConv2DErrorCase3(TestFunctionalConv2DError):
self.dilation = 1 self.dilation = 1
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NHWC" self.data_format = "NHWC"
...@@ -445,8 +407,6 @@ class TestFunctionalConv2DErrorCase4(TestFunctionalConv2DError): ...@@ -445,8 +407,6 @@ class TestFunctionalConv2DErrorCase4(TestFunctionalConv2DError):
self.dilation = 1 self.dilation = 1
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NCHW" self.data_format = "NCHW"
...@@ -460,23 +420,6 @@ class TestFunctionalConv2DErrorCase5(TestFunctionalConv2DError): ...@@ -460,23 +420,6 @@ class TestFunctionalConv2DErrorCase5(TestFunctionalConv2DError):
self.dilation = 1 self.dilation = 1
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NCHW"
class TestFunctionalConv2DErrorCase6(TestFunctionalConv2DError):
def setUp(self):
self.in_channels = 4
self.out_channels = 5
self.filter_shape = 3
self.padding = 0
self.stride = 1
self.dilation = 1
self.groups = 1
self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = "not_valid"
self.data_format = "NCHW" self.data_format = "NCHW"
...@@ -491,8 +434,6 @@ class TestFunctionalConv2DErrorCase7(TestFunctionalConv2DError): ...@@ -491,8 +434,6 @@ class TestFunctionalConv2DErrorCase7(TestFunctionalConv2DError):
self.dilation = 1 self.dilation = 1
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NCHW" self.data_format = "NCHW"
...@@ -506,8 +447,6 @@ class TestFunctionalConv2DErrorCase8(TestFunctionalConv2DError): ...@@ -506,8 +447,6 @@ class TestFunctionalConv2DErrorCase8(TestFunctionalConv2DError):
self.dilation = 1 self.dilation = 1
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "not_valid" self.data_format = "not_valid"
...@@ -521,8 +460,6 @@ class TestFunctionalConv2DErrorCase9(TestFunctionalConv2DError): ...@@ -521,8 +460,6 @@ class TestFunctionalConv2DErrorCase9(TestFunctionalConv2DError):
self.dilation = 1 self.dilation = 1
self.groups = 2 self.groups = 2
self.no_bias = False self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NCHW" self.data_format = "NCHW"
......
python/paddle/fluid/tests/unittests/test_functional_conv3d_transpose.py
浏览文件 @ d26ae9ad
...@@ -38,7 +38,6 @@ class TestFunctionalConv3DTranspose(TestCase): ...@@ -38,7 +38,6 @@ class TestFunctionalConv3DTranspose(TestCase):
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid" self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NDHWC" self.data_format = "NDHWC"
def prepare(self): def prepare(self):
...@@ -90,7 +89,6 @@ class TestFunctionalConv3DTranspose(TestCase): ...@@ -90,7 +89,6 @@ class TestFunctionalConv3DTranspose(TestCase):
param_attr=I.NumpyArrayInitializer(self.weight), param_attr=I.NumpyArrayInitializer(self.weight),
bias_attr=False bias_attr=False
if self.no_bias else I.NumpyArrayInitializer(self.bias), if self.no_bias else I.NumpyArrayInitializer(self.bias),
use_cudnn=self.use_cudnn,
act=self.act, act=self.act,
data_format=self.data_format) data_format=self.data_format)
exe = fluid.Executor(self.place) exe = fluid.Executor(self.place)
...@@ -115,7 +113,7 @@ class TestFunctionalConv3DTranspose(TestCase): ...@@ -115,7 +113,7 @@ class TestFunctionalConv3DTranspose(TestCase):
"weight", self.weight.shape, dtype=self.dtype) "weight", self.weight.shape, dtype=self.dtype)
if not self.no_bias: if not self.no_bias:
bias = fluid.data("bias", self.bias.shape, dtype=self.dtype) bias = fluid.data("bias", self.bias.shape, dtype=self.dtype)
y = F.conv3d_transpose( y = F.conv_transpose3d(
x, x,
weight, weight,
None if self.no_bias else bias, None if self.no_bias else bias,
...@@ -124,9 +122,9 @@ class TestFunctionalConv3DTranspose(TestCase): ...@@ -124,9 +122,9 @@ class TestFunctionalConv3DTranspose(TestCase):
stride=self.stride, stride=self.stride,
dilation=self.dilation, dilation=self.dilation,
groups=self.groups, groups=self.groups,
act=self.act, data_format=self.data_format)
data_format=self.data_format, if self.act == 'sigmoid':
use_cudnn=self.use_cudnn) y = F.sigmoid(y)
exe = fluid.Executor(self.place) exe = fluid.Executor(self.place)
exe.run(start) exe.run(start)
feed_dict = {"input": self.input, "weight": self.weight} feed_dict = {"input": self.input, "weight": self.weight}
...@@ -140,7 +138,7 @@ class TestFunctionalConv3DTranspose(TestCase): ...@@ -140,7 +138,7 @@ class TestFunctionalConv3DTranspose(TestCase):
x = dg.to_variable(self.input) x = dg.to_variable(self.input)
weight = dg.to_variable(self.weight) weight = dg.to_variable(self.weight)
bias = None if self.no_bias else dg.to_variable(self.bias) bias = None if self.no_bias else dg.to_variable(self.bias)
y = F.conv3d_transpose( y = F.conv_transpose3d(
x, x,
weight, weight,
bias, bias,
...@@ -148,10 +146,10 @@ class TestFunctionalConv3DTranspose(TestCase): ...@@ -148,10 +146,10 @@ class TestFunctionalConv3DTranspose(TestCase):
padding=self.padding, padding=self.padding,
stride=self.stride, stride=self.stride,
dilation=self.dilation, dilation=self.dilation,
act=self.act,
groups=self.groups, groups=self.groups,
data_format=self.data_format, data_format=self.data_format)
use_cudnn=self.use_cudnn) if self.act == 'sigmoid':
y = F.sigmoid(y)
out = y.numpy() out = y.numpy()
return out return out
...@@ -190,7 +188,6 @@ class TestFunctionalConv3DTransposeError(TestCase): ...@@ -190,7 +188,6 @@ class TestFunctionalConv3DTransposeError(TestCase):
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid" self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NDHWC" self.data_format = "NDHWC"
def test_exception(self): def test_exception(self):
...@@ -225,7 +222,7 @@ class TestFunctionalConv3DTransposeError(TestCase): ...@@ -225,7 +222,7 @@ class TestFunctionalConv3DTransposeError(TestCase):
"weight", self.weight_shape, dtype=self.dtype) "weight", self.weight_shape, dtype=self.dtype)
if not self.no_bias: if not self.no_bias:
bias = fluid.data("bias", self.bias_shape, dtype=self.dtype) bias = fluid.data("bias", self.bias_shape, dtype=self.dtype)
y = F.conv3d_transpose( y = F.conv_transpose3d(
x, x,
weight, weight,
None if self.no_bias else bias, None if self.no_bias else bias,
...@@ -234,9 +231,9 @@ class TestFunctionalConv3DTransposeError(TestCase): ...@@ -234,9 +231,9 @@ class TestFunctionalConv3DTransposeError(TestCase):
stride=self.stride, stride=self.stride,
dilation=self.dilation, dilation=self.dilation,
groups=self.groups, groups=self.groups,
act=self.act, data_format=self.data_format)
data_format=self.data_format, if self.act == 'sigmoid':
use_cudnn=self.use_cudnn) y = F.sigmoid(y)
class TestFunctionalConv3DTransposeCase2(TestFunctionalConv3DTranspose): class TestFunctionalConv3DTransposeCase2(TestFunctionalConv3DTranspose):
...@@ -250,7 +247,6 @@ class TestFunctionalConv3DTransposeCase2(TestFunctionalConv3DTranspose): ...@@ -250,7 +247,6 @@ class TestFunctionalConv3DTransposeCase2(TestFunctionalConv3DTranspose):
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid" self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NCDHW" self.data_format = "NCDHW"
...@@ -265,7 +261,6 @@ class TestFunctionalConv3DTransposeCase3(TestFunctionalConv3DTranspose): ...@@ -265,7 +261,6 @@ class TestFunctionalConv3DTransposeCase3(TestFunctionalConv3DTranspose):
self.groups = 2 self.groups = 2
self.no_bias = False self.no_bias = False
self.act = "sigmoid" self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NDHWC" self.data_format = "NDHWC"
...@@ -280,7 +275,6 @@ class TestFunctionalConv3DTransposeCase4(TestFunctionalConv3DTranspose): ...@@ -280,7 +275,6 @@ class TestFunctionalConv3DTransposeCase4(TestFunctionalConv3DTranspose):
self.groups = 2 self.groups = 2
self.no_bias = True self.no_bias = True
self.act = "sigmoid" self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NDHWC" self.data_format = "NDHWC"
...@@ -295,7 +289,6 @@ class TestFunctionalConv3DTransposeCase5(TestFunctionalConv3DTranspose): ...@@ -295,7 +289,6 @@ class TestFunctionalConv3DTransposeCase5(TestFunctionalConv3DTranspose):
self.groups = 2 self.groups = 2
self.no_bias = False self.no_bias = False
self.act = "sigmoid" self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NDHWC" self.data_format = "NDHWC"
...@@ -310,7 +303,6 @@ class TestFunctionalConv3DTransposeCase6(TestFunctionalConv3DTranspose): ...@@ -310,7 +303,6 @@ class TestFunctionalConv3DTransposeCase6(TestFunctionalConv3DTranspose):
self.groups = 4 self.groups = 4
self.no_bias = False self.no_bias = False
self.act = "sigmoid" self.act = "sigmoid"
self.use_cudnn = False
self.data_format = "NDHWC" self.data_format = "NDHWC"
...@@ -326,7 +318,6 @@ class TestFunctionalConv3DTransposeCase7(TestFunctionalConv3DTranspose): ...@@ -326,7 +318,6 @@ class TestFunctionalConv3DTransposeCase7(TestFunctionalConv3DTranspose):
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid" self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NCDHW" self.data_format = "NCDHW"
...@@ -341,7 +332,6 @@ class TestFunctionalConv3DTransposeCase8(TestFunctionalConv3DTranspose): ...@@ -341,7 +332,6 @@ class TestFunctionalConv3DTransposeCase8(TestFunctionalConv3DTranspose):
self.groups = 2 self.groups = 2
self.no_bias = False self.no_bias = False
self.act = "sigmoid" self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NDHWC" self.data_format = "NDHWC"
...@@ -356,7 +346,6 @@ class TestFunctionalConv3DTransposeCase9(TestFunctionalConv3DTranspose): ...@@ -356,7 +346,6 @@ class TestFunctionalConv3DTransposeCase9(TestFunctionalConv3DTranspose):
self.groups = 2 self.groups = 2
self.no_bias = False self.no_bias = False
self.act = "sigmoid" self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NCDHW" self.data_format = "NCDHW"
...@@ -371,7 +360,6 @@ class TestFunctionalConv3DTransposeCase10(TestFunctionalConv3DTranspose): ...@@ -371,7 +360,6 @@ class TestFunctionalConv3DTransposeCase10(TestFunctionalConv3DTranspose):
self.groups = 2 self.groups = 2
self.no_bias = False self.no_bias = False
self.act = "sigmoid" self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NCDHW" self.data_format = "NCDHW"
...@@ -386,7 +374,6 @@ class TestFunctionalConv3DTransposeCase11(TestFunctionalConv3DTranspose): ...@@ -386,7 +374,6 @@ class TestFunctionalConv3DTransposeCase11(TestFunctionalConv3DTranspose):
self.groups = 2 self.groups = 2
self.no_bias = False self.no_bias = False
self.act = "sigmoid" self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NCDHW" self.data_format = "NCDHW"
...@@ -402,7 +389,6 @@ class TestFunctionalConv3DTransposeErrorCase2( ...@@ -402,7 +389,6 @@ class TestFunctionalConv3DTransposeErrorCase2(
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid" self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NDHWC" self.data_format = "NDHWC"
...@@ -418,7 +404,6 @@ class TestFunctionalConv3DTransposeErrorCase3( ...@@ -418,7 +404,6 @@ class TestFunctionalConv3DTransposeErrorCase3(
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid" self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NDHWC" self.data_format = "NDHWC"
...@@ -434,7 +419,6 @@ class TestFunctionalConv3DTransposeErrorCase4( ...@@ -434,7 +419,6 @@ class TestFunctionalConv3DTransposeErrorCase4(
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid" self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NCDHW" self.data_format = "NCDHW"
...@@ -450,23 +434,6 @@ class TestFunctionalConv3DTransposeErrorCase5( ...@@ -450,23 +434,6 @@ class TestFunctionalConv3DTransposeErrorCase5(
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid" self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NCDHW"
class TestFunctionalConv3DTransposeErrorCase6(
TestFunctionalConv3DTransposeError):
def setUp(self):
self.in_channels = 4
self.out_channels = 5
self.filter_shape = 3
self.padding = 0
self.stride = 1
self.dilation = 1
self.groups = 1
self.no_bias = False
self.act = "sigmoid"
self.use_cudnn = "not_valid"
self.data_format = "NCDHW" self.data_format = "NCDHW"
...@@ -483,7 +450,6 @@ class TestFunctionalConv3DTransposeErrorCase7( ...@@ -483,7 +450,6 @@ class TestFunctionalConv3DTransposeErrorCase7(
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid" self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NCDHW" self.data_format = "NCDHW"
...@@ -499,7 +465,6 @@ class TestFunctionalConv3DTransposeErrorCase8( ...@@ -499,7 +465,6 @@ class TestFunctionalConv3DTransposeErrorCase8(
self.groups = 1 self.groups = 1
self.no_bias = False self.no_bias = False
self.act = "sigmoid" self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "not_valid" self.data_format = "not_valid"
...@@ -515,7 +480,6 @@ class TestFunctionalConv3DTransposeErrorCase9( ...@@ -515,7 +480,6 @@ class TestFunctionalConv3DTransposeErrorCase9(
self.groups = 2 self.groups = 2
self.no_bias = False self.no_bias = False
self.act = "sigmoid" self.act = "sigmoid"
self.use_cudnn = True
self.data_format = "NCDHW" self.data_format = "NCDHW"
......
python/paddle/nn/__init__.py
浏览文件 @ d26ae9ad
...@@ -94,9 +94,9 @@ from .layer.common import Dropout3D #DEFINE_ALIAS ...@@ -94,9 +94,9 @@ from .layer.common import Dropout3D #DEFINE_ALIAS
from .layer.pooling import AdaptiveAvgPool2d #DEFINE_ALIAS from .layer.pooling import AdaptiveAvgPool2d #DEFINE_ALIAS
from .layer.pooling import AdaptiveAvgPool3d #DEFINE_ALIAS from .layer.pooling import AdaptiveAvgPool3d #DEFINE_ALIAS
from .layer.conv import Conv2D #DEFINE_ALIAS from .layer.conv import Conv2D #DEFINE_ALIAS
from .layer.conv import Conv2DTranspose #DEFINE_ALIAS from .layer.conv import ConvTranspose2d #DEFINE_ALIAS
from .layer.conv import Conv3D #DEFINE_ALIAS from .layer.conv import Conv3D #DEFINE_ALIAS
from .layer.conv import Conv3DTranspose #DEFINE_ALIAS from .layer.conv import ConvTranspose3d #DEFINE_ALIAS
# from .layer.conv import TreeConv #DEFINE_ALIAS # from .layer.conv import TreeConv #DEFINE_ALIAS
# from .layer.conv import Conv1D #DEFINE_ALIAS # from .layer.conv import Conv1D #DEFINE_ALIAS
from .layer.extension import RowConv #DEFINE_ALIAS from .layer.extension import RowConv #DEFINE_ALIAS
......
python/paddle/nn/functional/__init__.py
浏览文件 @ d26ae9ad
...@@ -70,9 +70,9 @@ from .common import unfold #DEFINE_ALIAS ...@@ -70,9 +70,9 @@ from .common import unfold #DEFINE_ALIAS
from .common import assign #DEFINE_ALIAS from .common import assign #DEFINE_ALIAS
from .common import interpolate #DEFINE_ALIAS from .common import interpolate #DEFINE_ALIAS
from .conv import conv2d #DEFINE_ALIAS from .conv import conv2d #DEFINE_ALIAS
from .conv import conv2d_transpose #DEFINE_ALIAS from .conv import conv_transpose2d #DEFINE_ALIAS
from .conv import conv3d #DEFINE_ALIAS from .conv import conv3d #DEFINE_ALIAS
from .conv import conv3d_transpose #DEFINE_ALIAS from .conv import conv_transpose3d #DEFINE_ALIAS
from .extension import add_position_encoding #DEFINE_ALIAS from .extension import add_position_encoding #DEFINE_ALIAS
# from .extension import autoincreased_step_counter #DEFINE_ALIAS # from .extension import autoincreased_step_counter #DEFINE_ALIAS
from .extension import continuous_value_model #DEFINE_ALIAS from .extension import continuous_value_model #DEFINE_ALIAS
......
python/paddle/nn/functional/conv.py
浏览文件 @ d26ae9ad
...@@ -13,9 +13,10 @@ ...@@ -13,9 +13,10 @@
# limitations under the License. # limitations under the License.
from __future__ import print_function from __future__ import print_function
__all__ = ['conv2d', 'conv2d_transpose', 'conv3d', 'conv3d_transpose'] __all__ = ['conv2d', 'conv_transpose2d', 'conv3d', 'conv_transpose3d']
import numpy as np import numpy as np
from ...device import get_cudnn_version
from ...fluid.framework import Variable, in_dygraph_mode from ...fluid.framework import Variable, in_dygraph_mode
from ...fluid import core, dygraph_utils from ...fluid import core, dygraph_utils
from ...fluid.layers import nn, utils from ...fluid.layers import nn, utils
...@@ -323,21 +324,18 @@ def conv2d(input, ...@@ -323,21 +324,18 @@ def conv2d(input,
return out return out
def conv2d_transpose(input, def conv_transpose2d(x,
weight, weight,
bias=None, bias=None,
output_size=None,
padding=0,
stride=1, stride=1,
dilation=1, padding=0,
output_padding=0,
groups=1, groups=1,
use_cudnn=True, dilation=1,
act=None,
data_format='NCHW', data_format='NCHW',
output_size=None,
name=None): name=None):
""" """
:alias_main: paddle.nn.functional.conv2d_transpose
:alias: paddle.nn.functional.conv2d_transpose,paddle.nn.functional.conv.conv2d_transpose
The convolution2D transpose layer calculates the output based on the input, The convolution2D transpose layer calculates the output based on the input,
filter, and dilations, strides, paddings. Input(Input) and output(Output) filter, and dilations, strides, paddings. Input(Input) and output(Output)
...@@ -350,6 +348,7 @@ def conv2d_transpose(input, ...@@ -350,6 +348,7 @@ def conv2d_transpose(input,
If bias attribution and activation type are provided, bias is added to If bias attribution and activation type are provided, bias is added to
the output of the convolution, and the corresponding activation function the output of the convolution, and the corresponding activation function
is applied to the final result. is applied to the final result.
See more detail in :ref:`api_nn_conv_ConvTranspose2d` .
For each input :math:`X`, the equation is: For each input :math:`X`, the equation is:
...@@ -398,18 +397,15 @@ def conv2d_transpose(input, ...@@ -398,18 +397,15 @@ def conv2d_transpose(input,
conv2d_transpose can compute the kernel size automatically. conv2d_transpose can compute the kernel size automatically.
Args: Args:
input(Variable): 4-D Tensor with [N, C, H, W] or [N, H, W, C] format, x(Tensor): 4-D Tensor with [N, C, H, W] or [N, H, W, C] format,
whose data type is float32 or float64. whose data type is float32 or float64.
weight(Variable): The convolution kernel, a Tensor with shape [C, M/g, kH, kW], weight(Tensor): The convolution kernel, a Tensor with shape [C, M/g, kH, kW],
where M is the number of output channels(filters), g is the number of groups, where M is the number of output channels(filters), g is the number of groups,
kH is the height of the kernel, and kW is the width of the kernel. kH is the height of the kernel, and kW is the width of the kernel.
bias(Variable, optional): The bias, a Tensor with shape [M, ]. bias(Tensor, optional): The bias, a Tensor with shape [M, ].
output_size(int|tuple|list, optional): The output image size. If output size is a stride(int|list|tuple, optional): The stride size. It means the stride in transposed convolution.
tuple, it must contain two integers, (image_height, image_width). None if use If stride is a tuple, it must contain two integers, (stride_height, stride_width).
filter_size, padding, and stride to calculate output_size. Otherwise, stride_height = stride_width = stride. Default: stride = 1.
If output_size is specified, output_size and filter_size (weight)'s shape
should follow the formula above. Default: None. output_size and filter_size
should not be None at the same time.
padding(int|list|str|tuple, optional): The padding size. The padding argument effectively adds padding(int|list|str|tuple, optional): The padding size. The padding argument effectively adds
`dilation * (kernel - 1)` amount of zero-padding on both sides of input. If `padding` is a `dilation * (kernel - 1)` amount of zero-padding on both sides of input. If `padding` is a
string, either 'VALID' or 'SAME' supported, which is the padding algorithm. string, either 'VALID' or 'SAME' supported, which is the padding algorithm.
...@@ -421,10 +417,9 @@ def conv2d_transpose(input, ...@@ -421,10 +417,9 @@ def conv2d_transpose(input,
when `data_format` is `'NHWC'`, `padding` can be in the form when `data_format` is `'NHWC'`, `padding` can be in the form
`[[0,0], [pad_height_top, pad_height_bottom], [pad_width_left, pad_width_right], [0,0]]`. `[[0,0], [pad_height_top, pad_height_bottom], [pad_width_left, pad_width_right], [0,0]]`.
Default: padding = 0. Default: padding = 0.
stride(int|tuple, optional): The stride size. It means the stride in transposed convolution. output_padding(int|list|tuple, optional): Additional size added to one side
If stride is a tuple, it must contain two integers, (stride_height, stride_width). of each dimension in the output shape. Default: 0.
Otherwise, stride_height = stride_width = stride. Default: stride = 1. dilation(int|list|tuple, optional): The dilation size. It means the spacing between the kernel points.
dilation(int|tuple, optional): The dilation size. It means the spacing between the kernel points.
If dilation is a tuple, it must contain two integers, (dilation_height, dilation_width). If dilation is a tuple, it must contain two integers, (dilation_height, dilation_width).
Otherwise, dilation_height = dilation_width = dilation. Default: dilation = 1. Otherwise, dilation_height = dilation_width = dilation. Default: dilation = 1.
groups(int, optional): The groups number of the Conv2d transpose layer. Inspired by groups(int, optional): The groups number of the Conv2d transpose layer. Inspired by
...@@ -433,10 +428,12 @@ def conv2d_transpose(input, ...@@ -433,10 +428,12 @@ def conv2d_transpose(input,
first half of the input channels, while the second half of the first half of the input channels, while the second half of the
filters is only connected to the second half of the input channels. filters is only connected to the second half of the input channels.
Default: groups = 1. Default: groups = 1.
use_cudnn(bool, optional): Use cudnn kernel or not, it is valid only when the cudnn output_size(int|tuple|list, optional): The output image size. If output size is a
library is installed. Default: True. tuple, it must contain two integers, (image_height, image_width). None if use
act (str, optional): Activation type, if it is set to None, activation is not appended. filter_size, padding, and stride to calculate output_size.
Default: None. If output_size is specified, output_size and filter_size (weight)'s shape
should follow the formula above. Default: None. output_size and filter_size
should not be None at the same time.
data_format (str, optional): Specify the data format of the input, and the data format of the output data_format (str, optional): Specify the data format of the input, and the data format of the output
will be consistent with that of the input. An optional string from: `"NCHW"`, `"NHWC"`. will be consistent with that of the input. An optional string from: `"NCHW"`, `"NHWC"`.
The default is `"NCHW"`. When it is `"NCHW"`, the data is stored in the order of: The default is `"NCHW"`. When it is `"NCHW"`, the data is stored in the order of:
...@@ -446,20 +443,17 @@ def conv2d_transpose(input, ...@@ -446,20 +443,17 @@ def conv2d_transpose(input,
None by default. None by default.
Returns: Returns:
A Variable holding Tensor representing the conv2d_transpose, whose A Tensor representing the conv_transpose2d, whose
data type is the same with input and shape is (num_batches, channels, out_h, data type is the same with input and shape is (num_batches, channels, out_h,
out_w) or (num_batches, out_h, out_w, channels). If act is None, the tensor variable out_w) or (num_batches, out_h, out_w, channels). The tensor variable storing
storing the transposed convolution result, and if act is not None, the transposed convolution result.
tensor variable storing transposed convolution and non-linearity activation
result.
Raises: Raises:
ValueError: If the type of `use_cudnn` is not bool.
ValueError: If `data_format` is not "NCHW" or "NHWC". ValueError: If `data_format` is not "NCHW" or "NHWC".
ValueError: If `padding` is a string, but not "SAME" or "VALID". ValueError: If `padding` is a string, but not "SAME" or "VALID".
ValueError: If `padding` is a tuple, but the element corresponding to the input's batch size is not 0 ValueError: If `padding` is a tuple, but the element corresponding to the input's batch size is not 0
or the element corresponding to the input's channel is not 0. or the element corresponding to the input's channel is not 0.
ValueError: If `output_size` and filter_size are None at the same time. ValueError: If `output_size` and kernel_size are None at the same time.
ShapeError: If the input is not 4-D Tensor. ShapeError: If the input is not 4-D Tensor.
ShapeError: If the input's dimension size and filter's dimension size not equal. ShapeError: If the input's dimension size and filter's dimension size not equal.
ShapeError: If the dimension size of input minus the size of `stride` is not 2. ShapeError: If the dimension size of input minus the size of `stride` is not 2.
...@@ -469,28 +463,23 @@ def conv2d_transpose(input, ...@@ -469,28 +463,23 @@ def conv2d_transpose(input,
Examples: Examples:
.. code-block:: python .. code-block:: python
from paddle import fluid
import paddle.nn.functional as F
import paddle.fluid.dygraph as dg
import numpy as np import numpy as np
import paddle
import paddle.nn.functional as F
x = np.random.randn(2, 3, 8, 8).astype(np.float32) x = np.random.randn(2, 3, 8, 8).astype(np.float32)
w = np.random.randn(3, 6, 3, 3).astype(np.float32) w = np.random.randn(3, 6, 3, 3).astype(np.float32)
place = fluid.CPUPlace() paddle.disable_static()
with dg.guard(place): x_var = paddle.to_tensor(x)
x_var = dg.to_variable(x) w_var = paddle.to_tensor(w)
w_var = dg.to_variable(w) y_var = F.conv2d_transpose(x_var, w_var)
y_var = F.conv2d_transpose(x_var, w_var, act="relu")
y_np = y_var.numpy() y_np = y_var.numpy()
print(y_np.shape) print(y_np.shape)
# (2, 6, 10, 10) # (2, 6, 10, 10)
""" """
if not isinstance(use_cudnn, bool):
raise ValueError("Attr(use_cudnn) should be True or False. "
"Received Attr(use_cudnn): {}.".format(use_cudnn))
if data_format not in ['NCHW', 'NHWC']: if data_format not in ['NCHW', 'NHWC']:
raise ValueError( raise ValueError(
"Attr(data_format) of conv2d_transpose got wrong value: " "Attr(data_format) of conv2d_transpose got wrong value: "
...@@ -498,48 +487,65 @@ def conv2d_transpose(input, ...@@ -498,48 +487,65 @@ def conv2d_transpose(input,
data_format)) data_format))
channel_last = (data_format == "NHWC") channel_last = (data_format == "NHWC")
channel_dim = -1 if channel_last else 1 channel_dim = -1 if channel_last else 1
num_channels = input.shape[channel_dim] num_channels = x.shape[channel_dim]
if num_channels < 0: if num_channels < 0:
raise ValueError("The channel dimmention of the input({}) " raise ValueError("The channel dimmention of the input({}) "
"should be defined. Received: {}.".format( "should be defined. Received: {}.".format(
input.shape, num_channels)) x.shape, num_channels))
if num_channels % groups != 0: if num_channels % groups != 0:
raise ValueError( raise ValueError(
"the channel of input must be divisible by groups," "the channel of input must be divisible by groups,"
"received: the channel of input is {}, the shape of input is {}" "received: the channel of input is {}, the shape of input is {}"
", the groups is {}".format(num_channels, input.shape, groups)) ", the groups is {}".format(num_channels, x.shape, groups))
cudnn_version = get_cudnn_version()
use_cudnn = True if (core.is_compiled_with_cuda() and
cudnn_version is not None) else False
# update attrs # update attrs
padding, padding_algorithm = _update_padding_nd(padding, channel_last, 2) padding, padding_algorithm = _update_padding_nd(padding, channel_last, 2)
stride = utils.convert_to_list(stride, 2, 'stride') stride = utils.convert_to_list(stride, 2, 'stride')
dilation = utils.convert_to_list(dilation, 2, 'dilation') dilation = utils.convert_to_list(dilation, 2, 'dilation')
if output_size is None: if output_size is None:
output_size = [] output_size = []
elif isinstance(output_size, (list, tuple, int)): else:
if output_padding != 0:
raise ValueError('output_padding option is mutually exclusive with '
'output_size')
if isinstance(output_size, (list, tuple, int)):
output_size = utils.convert_to_list(output_size, 2, 'output_size') output_size = utils.convert_to_list(output_size, 2, 'output_size')
else: else:
raise ValueError("output_size should be int, or list, tuple of ints") raise ValueError(
"output_size should be int, or list, tuple of ints")
if output_padding == 0:
output_padding = []
else:
output_padding = utils.convert_to_list(output_padding, 2,
'output_padding')
op_type = 'conv2d_transpose' op_type = 'conv2d_transpose'
num_filters = weight.shape[1] num_filters = weight.shape[1]
if (num_channels == groups and num_filters == 1 and not use_cudnn): if (num_channels == groups and num_filters == 1):
op_type = 'depthwise_conv2d_transpose' op_type = 'depthwise_conv2d_transpose'
use_cudnn = False
if in_dygraph_mode(): if in_dygraph_mode():
attrs = ('output_size', output_size, 'strides', stride, 'paddings', attrs = ('output_padding', output_padding, 'output_size', output_size,
padding, 'padding_algorithm', padding_algorithm, 'dilations', 'strides', stride, 'paddings', padding, 'padding_algorithm',
dilation, 'groups', groups, 'use_cudnn', use_cudnn, padding_algorithm, 'dilations', dilation, 'groups', groups,
'data_format', data_format) 'use_cudnn', use_cudnn, 'data_format', data_format)
pre_bias = getattr(core.ops, op_type)(input, weight, *attrs) pre_bias = getattr(core.ops, op_type)(x, weight, *attrs)
if bias is not None: if bias is not None:
pre_act = nn.elementwise_add(pre_bias, bias, axis=channel_dim) out = nn.elementwise_add(pre_bias, bias, axis=channel_dim)
else: else:
pre_act = pre_bias out = pre_bias
out = dygraph_utils._append_activation_in_dygraph(
pre_act, act, use_cudnn=use_cudnn)
else: else:
inputs = {'Input': [input], 'Filter': [weight]} inputs = {'Input': [x], 'Filter': [weight]}
attrs = { attrs = {
'output_padding': output_padding,
'output_size': output_size, 'output_size': output_size,
'strides': stride, 'strides': stride,
'paddings': padding, 'paddings': padding,
...@@ -549,20 +555,19 @@ def conv2d_transpose(input, ...@@ -549,20 +555,19 @@ def conv2d_transpose(input,
'use_cudnn': use_cudnn, 'use_cudnn': use_cudnn,
'data_format': data_format 'data_format': data_format
} }
check_variable_and_dtype(input, 'input', check_variable_and_dtype(x, 'x', ['float16', 'float32', 'float64'],
['float16', 'float32', 'float64'],
'conv2d_transpose') 'conv2d_transpose')
helper = LayerHelper(op_type, **locals()) helper = LayerHelper(op_type, **locals())
dtype = helper.input_dtype() pre_bias = helper.create_variable_for_type_inference(x.dtype)
pre_bias = helper.create_variable_for_type_inference(dtype)
outputs = {"Output": [pre_bias]} outputs = {"Output": [pre_bias]}
helper.append_op( helper.append_op(
type=op_type, inputs=inputs, outputs=outputs, attrs=attrs) type=op_type, inputs=inputs, outputs=outputs, attrs=attrs)
if bias is not None: if bias is not None:
pre_act = nn.elementwise_add(pre_bias, bias, axis=channel_dim) out = nn.elementwise_add(pre_bias, bias, axis=channel_dim)
else: else:
pre_act = pre_bias out = pre_bias
out = helper.append_activation(pre_act)
return out return out
...@@ -782,23 +787,19 @@ def conv3d(input, ...@@ -782,23 +787,19 @@ def conv3d(input,
return out return out
def conv3d_transpose(input, def conv_transpose3d(x,
weight, weight,
bias=None, bias=None,
output_size=None,
padding=0,
stride=1, stride=1,
dilation=1, padding=0,
output_padding=0,
groups=1, groups=1,
use_cudnn=True, dilation=1,
act=None,
data_format='NCDHW', data_format='NCDHW',
output_size=None,
name=None): name=None):
""" """
:alias_main: paddle.nn.functional.conv3d_transpose The convolution3d transpose layer calculates the output based on the input,
:alias: paddle.nn.functional.conv3d_transpose,paddle.nn.functional.conv.conv3d_transpose
The convolution3D transpose layer calculates the output based on the input,
filter, and dilations, strides, paddings. Input(Input) and output(Output) filter, and dilations, strides, paddings. Input(Input) and output(Output)
are in NCDHW or NDHWC format. Where N is batch size, C is the number of channels, are in NCDHW or NDHWC format. Where N is batch size, C is the number of channels,
D is the depth of the feature, H is the height of the feature, and W D is the depth of the feature, H is the height of the feature, and W
...@@ -809,6 +810,7 @@ def conv3d_transpose(input, ...@@ -809,6 +810,7 @@ def conv3d_transpose(input,
If bias attribution and activation type are provided, bias is added to If bias attribution and activation type are provided, bias is added to
the output of the convolution, and the corresponding activation function the output of the convolution, and the corresponding activation function
is applied to the final result. is applied to the final result.
See more detail in :ref:`api_nn_conv_ConvTranspose3d` .
For each input :math:`X`, the equation is: For each input :math:`X`, the equation is:
...@@ -861,17 +863,16 @@ def conv3d_transpose(input, ...@@ -861,17 +863,16 @@ def conv3d_transpose(input,
conv3d_transpose can compute the kernel size automatically. conv3d_transpose can compute the kernel size automatically.
Args: Args:
input(Variable): The input is 5-D Tensor with shape [N, C, D, H, W] or [N, D, H, W, C], the data type x(Tensor): The input is 5-D Tensor with shape [N, C, D, H, W] or [N, D, H, W, C], the data type
of input is float32 or float64. of input is float32 or float64.
weight (Variable): The convolution kernel, a Tensor with shape [C, M/g, kD, kH, kW], weight (Tensor): The convolution kernel, a Tensor with shape [C, M/g, kD, kH, kW],
where M is the number of filters(output channels), g is the number of groups, where M is the number of filters(output channels), g is the number of groups,
kD, kH, kW are the filter's depth, height and width respectively. kD, kH, kW are the filter's depth, height and width respectively.
bias (Variable, optional): The bias, a Tensor of shape [M, ]. bias (Tensor, optional): The bias, a Tensor of shape [M, ].
output_size(int|tuple, optional): The output image size. If output size is a stride(int|list|tuple, optional): The stride size. It means the stride in transposed convolution.
tuple, it must contain three integers, (image_depth, image_height, image_width). This If stride is a tuple, it must contain three integers, (stride_depth, stride_height,
parameter only works when filter_size is None. If output_size and filter_size are stride_width). Otherwise, stride_depth = stride_height = stride_width = stride.
specified at the same time, They should follow the formula above. Default: None. Default: stride = 1.
Output_size and filter_size should not be None at the same time.
padding(int|list|str|tuple, optional): The padding size. The padding argument effectively padding(int|list|str|tuple, optional): The padding size. The padding argument effectively
adds `dilation * (kernel - 1)` amount of zero-padding on both sides of input. If `padding` is a string, adds `dilation * (kernel - 1)` amount of zero-padding on both sides of input. If `padding` is a string,
either 'VALID' or 'SAME' supported, which is the padding algorithm. If `padding` either 'VALID' or 'SAME' supported, which is the padding algorithm. If `padding`
...@@ -882,11 +883,9 @@ def conv3d_transpose(input, ...@@ -882,11 +883,9 @@ def conv3d_transpose(input,
when `data_format` is `'NDHWC'`, `padding` can be in the form when `data_format` is `'NDHWC'`, `padding` can be in the form
`[[0,0], [pad_depth_front, pad_depth_back], [pad_height_top, pad_height_bottom], [pad_width_left, pad_width_right], [0,0]]`. `[[0,0], [pad_depth_front, pad_depth_back], [pad_height_top, pad_height_bottom], [pad_width_left, pad_width_right], [0,0]]`.
Default: padding = 0. Default: padding = 0.
stride(int|tuple, optional): The stride size. It means the stride in transposed convolution. output_padding(int|list|tuple, optional): Additional size added to one side
If stride is a tuple, it must contain three integers, (stride_depth, stride_height, of each dimension in the output shape. Default: 0.
stride_width). Otherwise, stride_depth = stride_height = stride_width = stride. dilation(int|list|tuple, optional): The dilation size. It means the spacing between the kernel points.
Default: stride = 1.
dilation(int|tuple, optional): The dilation size. It means the spacing between the kernel points.
If dilation is a tuple, it must contain three integers, (dilation_depth, dilation_height, If dilation is a tuple, it must contain three integers, (dilation_depth, dilation_height,
dilation_width). Otherwise, dilation_depth = dilation_height = dilation_width = dilation. dilation_width). Otherwise, dilation_depth = dilation_height = dilation_width = dilation.
Default: dilation = 1. Default: dilation = 1.
...@@ -896,32 +895,32 @@ def conv3d_transpose(input, ...@@ -896,32 +895,32 @@ def conv3d_transpose(input,
first half of the input channels, while the second half of the first half of the input channels, while the second half of the
filters is only connected to the second half of the input channels. filters is only connected to the second half of the input channels.
Default: groups=1 Default: groups=1
use_cudnn(bool, optional): Use cudnn kernel or not, it is valid only when the cudnn
library is installed. Default: True
act (str, optional): Activation type, if it is set to None, activation is not appended.
Default: None.
data_format (str, optional): Specify the data format of the input, and the data format of the output data_format (str, optional): Specify the data format of the input, and the data format of the output
will be consistent with that of the input. An optional string from: `"NCHW"`, `"NHWC"`. will be consistent with that of the input. An optional string from: `"NCHW"`, `"NHWC"`.
The default is `"NCHW"`. When it is `"NCHW"`, the data is stored in the order of: The default is `"NCHW"`. When it is `"NCHW"`, the data is stored in the order of:
`[batch_size, input_channels, input_height, input_width]`. `[batch_size, input_channels, input_height, input_width]`.
output_size(int|list|tuple, optional): The output image size. If output size is a
tuple, it must contain three integers, (image_depth, image_height, image_width). This
parameter only works when filter_size is None. If output_size and filter_size are
specified at the same time, They should follow the formula above. Default: None.
Output_size and filter_size should not be None at the same time.
name(str, optional): For detailed information, please refer name(str, optional): For detailed information, please refer
to :ref:`api_guide_Name`. Usually name is no need to set and to :ref:`api_guide_Name`. Usually name is no need to set and
None by default. None by default.
Returns: Returns:
A Variable holding Tensor representing the conv3d_transpose, whose data A Tensor representing the conv_transpose3d, whose data
type is the same with input and shape is (num_batches, channels, out_d, out_h, type is the same with input and shape is (num_batches, channels, out_d, out_h,
out_w) or (num_batches, out_d, out_h, out_w, channels). If act is None, the tensor out_w) or (num_batches, out_d, out_h, out_w, channels). If act is None, the tensor
variable storing the transposed convolution result, and if act is not None, the tensor variable storing the transposed convolution result, and if act is not None, the tensor
variable storing transposed convolution and non-linearity activation result. variable storing transposed convolution and non-linearity activation result.
Raises: Raises:
ValueError: If the type of `use_cudnn` is not bool.
ValueError: If `data_format` is not "NCDHW" or "NDHWC". ValueError: If `data_format` is not "NCDHW" or "NDHWC".
ValueError: If `padding` is a string, but not "SAME" or "VALID". ValueError: If `padding` is a string, but not "SAME" or "VALID".
ValueError: If `padding` is a tuple, but the element corresponding to the input's batch size is not 0 ValueError: If `padding` is a tuple, but the element corresponding to the input's batch size is not 0
or the element corresponding to the input's channel is not 0. or the element corresponding to the input's channel is not 0.
ValueError: If `output_size` and filter_size are None at the same time. ValueError: If `output_size` and kernel_size are None at the same time.
ShapeError: If the input is not 5-D Tensor. ShapeError: If the input is not 5-D Tensor.
ShapeError: If the input's dimension size and filter's dimension size not equal. ShapeError: If the input's dimension size and filter's dimension size not equal.
ShapeError: If the dimension size of input minus the size of `stride` is not 2. ShapeError: If the dimension size of input minus the size of `stride` is not 2.
...@@ -931,28 +930,25 @@ def conv3d_transpose(input, ...@@ -931,28 +930,25 @@ def conv3d_transpose(input,
Examples: Examples:
.. code-block:: python .. code-block:: python
from paddle import fluid
import paddle.nn.functional as F
import paddle.fluid.dygraph as dg
import numpy as np import numpy as np
import paddle
import paddle.nn.functional as F
x = np.random.randn(2, 3, 8, 8, 8).astype(np.float32) x = np.random.randn(2, 3, 8, 8, 8).astype(np.float32)
w = np.random.randn(3, 6, 3, 3, 3).astype(np.float32) w = np.random.randn(3, 6, 3, 3, 3).astype(np.float32)
place = fluid.CPUPlace() paddle.disable_static()
with dg.guard(place):
x_var = dg.to_variable(x) x_var = paddle.to_tensor(x)
w_var = dg.to_variable(w) w_var = paddle.to_tensor(w)
y_var = F.conv3d_transpose(x_var, w_var, act="relu") y_var = F.conv_transpose3d(x_var, w_var)
y_np = y_var.numpy() y_np = y_var.numpy()
print(y_np.shape) print(y_np.shape)
# (2, 6, 10, 10, 10) # (2, 6, 10, 10, 10)
""" """
# entry checks # entry checks
if not isinstance(use_cudnn, bool):
raise ValueError("Attr(use_cudnn) should be True or False. "
"Received Attr(use_cudnn): {}.".format(use_cudnn))
if data_format not in ["NCDHW", "NDHWC"]: if data_format not in ["NCDHW", "NDHWC"]:
raise ValueError( raise ValueError(
"Attr(data_format) should be 'NCDHW' or 'NDHWC'. Received " "Attr(data_format) should be 'NCDHW' or 'NDHWC'. Received "
...@@ -960,12 +956,12 @@ def conv3d_transpose(input, ...@@ -960,12 +956,12 @@ def conv3d_transpose(input,
channel_last = (data_format == "NDHWC") channel_last = (data_format == "NDHWC")
channel_dim = -1 if channel_last else 1 channel_dim = -1 if channel_last else 1
num_channels = input.shape[channel_dim] num_channels = x.shape[channel_dim]
num_filters = weight.shape[1] num_filters = weight.shape[1]
if num_channels < 0: if num_channels < 0:
raise ValueError( raise ValueError(
"The channel dimmention of the input({}) should be defined. " "The channel dimmention of the input({}) should be defined. "
"Received: {}.".format(input.shape, num_channels)) "Received: {}.".format(x.shape, num_channels))
if num_channels % groups != 0: if num_channels % groups != 0:
raise ValueError( raise ValueError(
"The number of input channels must be divisible by Attr(groups). " "The number of input channels must be divisible by Attr(groups). "
...@@ -977,29 +973,45 @@ def conv3d_transpose(input, ...@@ -977,29 +973,45 @@ def conv3d_transpose(input,
dilation = utils.convert_to_list(dilation, 3, 'dilation') dilation = utils.convert_to_list(dilation, 3, 'dilation')
if output_size is None: if output_size is None:
output_size = [] output_size = []
elif isinstance(output_size, (list, tuple, int)): else:
if output_padding != 0:
raise ValueError('output_padding option is mutually exclusive with '
'output_size')
if isinstance(output_size, (list, tuple, int)):
output_size = utils.convert_to_list(output_size, 3, 'output_size') output_size = utils.convert_to_list(output_size, 3, 'output_size')
else: else:
raise ValueError("output_size should be int, or list, tuple of ints") raise ValueError(
"output_size should be int, or list, tuple of ints")
if output_padding == 0:
output_padding = []
else:
output_padding = utils.convert_to_list(output_padding, 3,
'output_padding')
cudnn_version = get_cudnn_version()
#TODO(LielinJiang): whether to use cudnn according to the version of cudnn
use_cudnn = True if (core.is_compiled_with_cuda() and
cudnn_version is not None) else False
op_type = 'conv3d_transpose' op_type = 'conv3d_transpose'
data_format_ = "NHWC" if channel_last else "NCHW" data_format_ = "NHWC" if channel_last else "NCHW"
if in_dygraph_mode(): if in_dygraph_mode():
attrs = ('output_size', output_size, 'paddings', padding, attrs = ('output_padding', output_padding, 'output_size', output_size,
"padding_algorithm", padding_algorithm, 'strides', stride, 'paddings', padding, "padding_algorithm", padding_algorithm,
'dilations', dilation, 'groups', groups, 'use_cudnn', 'strides', stride, 'dilations', dilation, 'groups', groups,
use_cudnn, "data_format", data_format_) 'use_cudnn', use_cudnn, "data_format", data_format_)
pre_bias = getattr(core.ops, op_type)(input, weight, *attrs) pre_bias = getattr(core.ops, op_type)(x, weight, *attrs)
if bias is not None: if bias is not None:
pre_act = nn.elementwise_add(pre_bias, bias, axis=channel_dim) out = nn.elementwise_add(pre_bias, bias, axis=channel_dim)
else: else:
pre_act = pre_bias out = pre_bias
out = dygraph_utils._append_activation_in_dygraph(
pre_act, act, use_cudnn=use_cudnn)
else: else:
inputs = {'Input': [input], 'Filter': [weight]} inputs = {'Input': [x], 'Filter': [weight]}
attrs = { attrs = {
'output_padding': output_padding,
'output_size': output_size, 'output_size': output_size,
'paddings': padding, 'paddings': padding,
"padding_algorithm": padding_algorithm, "padding_algorithm": padding_algorithm,
...@@ -1010,19 +1022,17 @@ def conv3d_transpose(input, ...@@ -1010,19 +1022,17 @@ def conv3d_transpose(input,
"data_format": data_format_ "data_format": data_format_
} }
helper = LayerHelper(op_type, **locals()) helper = LayerHelper(op_type, **locals())
dtype = helper.input_dtype() check_variable_and_dtype(x, 'x', ['float16', 'float32', 'float64'],
check_variable_and_dtype(input, 'input', 'conv3d')
['float16', 'float32', 'float64'], 'conv3d')
pre_bias = helper.create_variable_for_type_inference(dtype) pre_bias = helper.create_variable_for_type_inference(x.dtype)
outputs = {"Output": [pre_bias]} outputs = {"Output": [pre_bias]}
helper.append_op( helper.append_op(
type=op_type, inputs=inputs, outputs=outputs, attrs=attrs) type=op_type, inputs=inputs, outputs=outputs, attrs=attrs)
if bias is not None: if bias is not None:
pre_act = nn.elementwise_add(pre_bias, bias, axis=channel_dim) out = nn.elementwise_add(pre_bias, bias, axis=channel_dim)
else: else:
pre_act = pre_bias out = pre_bias
out = helper.append_activation(pre_act)
return out return out
python/paddle/nn/layer/__init__.py
浏览文件 @ d26ae9ad
...@@ -58,9 +58,9 @@ from .common import Dropout3D #DEFINE_ALIAS ...@@ -58,9 +58,9 @@ from .common import Dropout3D #DEFINE_ALIAS
from .pooling import AdaptiveAvgPool2d #DEFINE_ALIAS from .pooling import AdaptiveAvgPool2d #DEFINE_ALIAS
from .pooling import AdaptiveAvgPool3d #DEFINE_ALIAS from .pooling import AdaptiveAvgPool3d #DEFINE_ALIAS
from .conv import Conv2D #DEFINE_ALIAS from .conv import Conv2D #DEFINE_ALIAS
from .conv import Conv2DTranspose #DEFINE_ALIAS from .conv import ConvTranspose2d #DEFINE_ALIAS
from .conv import Conv3D #DEFINE_ALIAS from .conv import Conv3D #DEFINE_ALIAS
from .conv import Conv3DTranspose #DEFINE_ALIAS from .conv import ConvTranspose3d #DEFINE_ALIAS
# from .conv import TreeConv #DEFINE_ALIAS # from .conv import TreeConv #DEFINE_ALIAS
# from .conv import Conv1D #DEFINE_ALIAS # from .conv import Conv1D #DEFINE_ALIAS
from .extension import RowConv #DEFINE_ALIAS from .extension import RowConv #DEFINE_ALIAS
......
python/paddle/nn/layer/conv.py
浏览文件 @ d26ae9ad
...@@ -16,9 +16,9 @@ ...@@ -16,9 +16,9 @@
__all__ = [ __all__ = [
'Conv2D', 'Conv2D',
'Conv2DTranspose', 'ConvTranspose2d',
'Conv3D', 'Conv3D',
'Conv3DTranspose', 'ConvTranspose3d',
# 'TreeConv', # 'TreeConv',
# 'Conv1D' # 'Conv1D'
] ]
...@@ -38,6 +38,51 @@ def _get_default_param_initializer(num_channels, filter_size): ...@@ -38,6 +38,51 @@ def _get_default_param_initializer(num_channels, filter_size):
return Normal(0.0, std, 0) return Normal(0.0, std, 0)
class _ConvNd(layers.Layer):
def __init__(self,
in_channels,
out_channels,
kernel_size,
transposed,
dims,
stride=1,
padding=0,
padding_mode='zeros',
output_padding=0,
dilation=1,
groups=1,
weight_attr=None,
bias_attr=None,
data_format="NCHW"):
super(_ConvNd, self).__init__()
assert weight_attr is not False, "weight_attr should not be False in Conv."
self._param_attr = weight_attr
self._bias_attr = bias_attr
self._groups = groups
self._in_channels = in_channels
self._out_channels = out_channels
self._data_format = data_format
self._stride = utils.convert_to_list(stride, dims, 'stride')
self._dilation = utils.convert_to_list(dilation, dims, 'dilation')
self._kernel_size = utils.convert_to_list(kernel_size, dims,
'kernel_size')
self._padding = padding
self.output_padding = output_padding
if transposed:
filter_shape = [self._in_channels, out_channels // groups
] + self._kernel_size
else:
filter_shape = [out_channels, in_channels // groups
] + self._kernel_size
self.weight = self.create_parameter(
shape=filter_shape, attr=self._param_attr)
self.bias = self.create_parameter(
attr=self._bias_attr, shape=[self._out_channels], is_bias=True)
class Conv2D(layers.Layer): class Conv2D(layers.Layer):
""" """
:alias_main: paddle.nn.Conv2D :alias_main: paddle.nn.Conv2D
...@@ -236,12 +281,9 @@ class Conv2D(layers.Layer): ...@@ -236,12 +281,9 @@ class Conv2D(layers.Layer):
return out return out
class Conv2DTranspose(layers.Layer): class ConvTranspose2d(_ConvNd):
""" """
:alias_main: paddle.nn.Conv2DTranspose This interface is used to construct a callable object of the ``ConvTranspose2d`` class.
:alias: paddle.nn.Conv2DTranspose,paddle.nn.layer.Conv2DTranspose,paddle.nn.layer.conv.Conv2DTranspose
This interface is used to construct a callable object of the ``Conv2DTranspose`` class.
For more details, refer to code examples. For more details, refer to code examples.
The convolution2D transpose layer calculates the output based on the input, The convolution2D transpose layer calculates the output based on the input,
filter, and dilations, strides, paddings. Input and output filter, and dilations, strides, paddings. Input and output
...@@ -294,17 +336,13 @@ class Conv2DTranspose(layers.Layer): ...@@ -294,17 +336,13 @@ class Conv2DTranspose(layers.Layer):
W_{out} &\in [ W^\prime_{out}, W^\prime_{out} + strides[1] ) W_{out} &\in [ W^\prime_{out}, W^\prime_{out} + strides[1] )
Parameters: Parameters:
num_channels(int): The number of channels in the input image. in_channels(int): The number of channels in the input image.
num_filters(int): The number of the filter. It is as same as the output out_channels(int): The number of channels produced by the convolution.
feature map. kernel_size(int|list|uple): The kernel size. If kernel_size is a tuple,
filter_size(int or tuple): The filter size. If filter_size is a tuple, it must contain two integers, (kernel_size_H, kernel_size_W).
it must contain two integers, (filter_size_H, filter_size_W). Otherwise, the kernel will be a square.
Otherwise, the filter will be a square. output_padding(int|list|tuple, optional): Additional size added to one side
output_size(int or tuple, optional): The output image size. If output size is a of each dimension in the output shape. Default: 0.
tuple, it must contain two integers, (image_H, image_W). None if use
filter_size, padding, and stride to calculate output_size.
if output_size and filter_size are specified at the same time, They
should follow the formula above. Default: None.
padding(int|str|tuple|list, optional): The padding size. Padding coule be in one of the following forms. padding(int|str|tuple|list, optional): The padding size. Padding coule be in one of the following forms.
1. a string in ['valid', 'same']. 1. a string in ['valid', 'same'].
2. an int, which means each spartial dimension(depth, height, width) is zero paded by size of `padding` on both sides 2. an int, which means each spartial dimension(depth, height, width) is zero paded by size of `padding` on both sides
...@@ -312,10 +350,10 @@ class Conv2DTranspose(layers.Layer): ...@@ -312,10 +350,10 @@ class Conv2DTranspose(layers.Layer):
4. a list[int] or tuple[int] whose length is 2 * number of spartial dimensions. It has the form [pad_before, pad_after, pad_before, pad_after, ...] for all spartial dimensions. 4. a list[int] or tuple[int] whose length is 2 * number of spartial dimensions. It has the form [pad_before, pad_after, pad_before, pad_after, ...] for all spartial dimensions.
5. a list or tuple of pairs of ints. It has the form [[pad_before, pad_after], [pad_before, pad_after], ...]. Note that, the batch dimension and channel dimension are also included. Each pair of integers correspond to the amount of padding for a dimension of the input. Padding in batch dimension and channel dimension should be [0, 0] or (0, 0). 5. a list or tuple of pairs of ints. It has the form [[pad_before, pad_after], [pad_before, pad_after], ...]. Note that, the batch dimension and channel dimension are also included. Each pair of integers correspond to the amount of padding for a dimension of the input. Padding in batch dimension and channel dimension should be [0, 0] or (0, 0).
The default value is 0. The default value is 0.
stride(int or tuple, optional): The stride size. If stride is a tuple, it must stride(int|list|tuple, optional): The stride size. If stride is a tuple, it must
contain two integers, (stride_H, stride_W). Otherwise, the contain two integers, (stride_H, stride_W). Otherwise, the
stride_H = stride_W = stride. Default: 1. stride_H = stride_W = stride. Default: 1.
dilation(int or tuple, optional): The dilation size. If dilation is a tuple, it must dilation(int|list|tuple, optional): The dilation size. If dilation is a tuple, it must
contain two integers, (dilation_H, dilation_W). Otherwise, the contain two integers, (dilation_H, dilation_W). Otherwise, the
dilation_H = dilation_W = dilation. Default: 1. dilation_H = dilation_W = dilation. Default: 1.
groups(int, optional): The groups number of the Conv2d transpose layer. Inspired by groups(int, optional): The groups number of the Conv2d transpose layer. Inspired by
...@@ -324,44 +362,48 @@ class Conv2DTranspose(layers.Layer): ...@@ -324,44 +362,48 @@ class Conv2DTranspose(layers.Layer):
first half of the input channels, while the second half of the first half of the input channels, while the second half of the
filters is only connected to the second half of the input channels. filters is only connected to the second half of the input channels.
Default: 1. Default: 1.
param_attr (ParamAttr, optional): The parameter attribute for learnable weights(Parameter) weight_attr (ParamAttr, optional): The parameter attribute for learnable weights(Parameter)
of conv2d_transpose. If it is set to None or one attribute of ParamAttr, conv2d_transpose of conv2d_transpose. If it is set to None or one attribute of ParamAttr, conv2d_transpose
will create ParamAttr as param_attr. If the Initializer of the param_attr will create ParamAttr as param_attr. If the Initializer of the param_attr
is not set, the parameter is initialized with Xavier. Default: None. is not set, the parameter is initialized with Xavier. Default: None.
bias_attr (ParamAttr or bool, optional): The attribute for the bias of conv2d_transpose. bias_attr (ParamAttr|bool, optional): The attribute for the bias of conv2d_transpose.
If it is set to False, no bias will be added to the output units. If it is set to False, no bias will be added to the output units.
If it is set to None or one attribute of ParamAttr, conv2d_transpose If it is set to None or one attribute of ParamAttr, conv2d_transpose
will create ParamAttr as bias_attr. If the Initializer of the bias_attr will create ParamAttr as bias_attr. If the Initializer of the bias_attr
is not set, the bias is initialized zero. Default: None. is not set, the bias is initialized zero. Default: None.
use_cudnn(bool, optional): Use cudnn kernel or not, it is valid only when the cudnn
library is installed. Default: True.
act (str, optional): Activation type, if it is set to None, activation is not appended.
Default: None.
data_format (str, optional): Data format that specifies the layout of input. data_format (str, optional): Data format that specifies the layout of input.
It can be "NCHW" or "NHWC". Default: "NCHW". It can be "NCHW" or "NHWC". Default: "NCHW".
dtype (str, optional): Data type, it can be "float32" or "float64". Default: "float32".
Attribute: Attribute:
**weight** (Parameter): the learnable weights of filters of this layer. **weight** (Parameter): the learnable weights of filters of this layer.
**bias** (Parameter or None): the learnable bias of this layer. **bias** (Parameter or None): the learnable bias of this layer.
Returns: Shape:
None - x: :math:`(N, C_{in}, H_{in}, W_{in})`
- output: :math:`(N, C_{out}, H_{out}, W_{out})`
Where
.. math::
H^\prime_{out} &= (H_{in} - 1) * strides[0] - 2 * paddings[0] + dilations[0] * (kernel_size[0] - 1) + 1 \\\\
W^\prime_{out} &= (W_{in} - 1) * strides[1] - 2 * paddings[1] + dilations[1] * (kernel_size[1] - 1) + 1 \\\\
Examples: Examples:
.. code-block:: python .. code-block:: python
import numpy as np import numpy as np
from paddle import fluid import paddle
import paddle.fluid.dygraph as dg import paddle.nn as nn
from paddle import nn
x = np.random.uniform(-1, 1, (2, 4, 8, 8)).astype('float32') x = np.random.uniform(-1, 1, (2, 4, 8, 8)).astype('float32')
place = fluid.CPUPlace()
with dg.guard(place): paddle.disable_static()
x_var = dg.to_variable(x)
conv = nn.Conv2DTranspose(4, 6, (3, 3)) x_var = paddle.to_tensor(x)
conv = nn.ConvTranspose2d(4, 6, (3, 3))
y_var = conv(x_var) y_var = conv(x_var)
y_np = y_var.numpy() y_np = y_var.numpy()
print(y_np.shape) print(y_np.shape)
...@@ -370,67 +412,48 @@ class Conv2DTranspose(layers.Layer): ...@@ -370,67 +412,48 @@ class Conv2DTranspose(layers.Layer):
""" """
def __init__(self, def __init__(self,
num_channels, in_channels,
num_filters, out_channels,
filter_size, kernel_size,
output_size=None,
padding=0,
stride=1, stride=1,
padding=0,
output_padding=0,
dilation=1, dilation=1,
groups=1, groups=1,
param_attr=None, weight_attr=None,
bias_attr=None, bias_attr=None,
use_cudnn=True, data_format="NCHW"):
act=None, super(ConvTranspose2d, self).__init__(
data_format="NCHW", in_channels,
dtype='float32'): out_channels,
super(Conv2DTranspose, self).__init__() kernel_size,
assert param_attr is not False, "param_attr should not be False in conv2d_transpose." True,
self._param_attr = param_attr 2,
self._bias_attr = bias_attr stride=stride,
self._act = act padding=padding,
self._groups = groups dilation=dilation,
self._num_channels = num_channels output_padding=output_padding,
self._num_filters = num_filters groups=groups,
self._use_cudnn = use_cudnn weight_attr=weight_attr,
self._data_format = data_format bias_attr=bias_attr,
self._dtype = dtype data_format=data_format)
self._stride = utils.convert_to_list(stride, 2, 'stride') def forward(self, x, output_size=None):
self._dilation = utils.convert_to_list(dilation, 2, 'dilation')
self._filter_size = utils.convert_to_list(filter_size, 2, 'filter_size')
if output_size is None: if output_size is None:
self._output_size = output_size output_padding = self.output_padding
elif isinstance(output_size, (list, tuple, int)):
self._output_size = utils.convert_to_list(output_size, 2,
'output_size')
else: else:
raise ValueError( output_padding = 0
"output_size should be int, ot list[int] or tuple[int]")
self._padding = padding
filter_shape = [self._num_channels, num_filters // groups
] + self._filter_size
self.weight = self.create_parameter(
dtype=self._dtype, shape=filter_shape, attr=self._param_attr)
self.bias = self.create_parameter(
attr=self._bias_attr,
shape=[self._num_filters],
dtype=self._dtype,
is_bias=True)
def forward(self, input): out = F.conv_transpose2d(
out = F.conv2d_transpose( x,
input,
self.weight, self.weight,
bias=self.bias, bias=self.bias,
output_size=self._output_size,
padding=self._padding, padding=self._padding,
output_padding=output_padding,
stride=self._stride, stride=self._stride,
dilation=self._dilation, dilation=self._dilation,
groups=self._groups, groups=self._groups,
use_cudnn=self._use_cudnn, output_size=output_size,
act=self._act,
data_format=self._data_format) data_format=self._data_format)
return out return out
...@@ -628,10 +651,8 @@ class Conv3D(layers.Layer): ...@@ -628,10 +651,8 @@ class Conv3D(layers.Layer):
return out return out
class Conv3DTranspose(layers.Layer): class ConvTranspose3d(_ConvNd):
""" """
:alias_main: paddle.nn.Conv3DTranspose
:alias: paddle.nn.Conv3DTranspose,paddle.nn.layer.Conv3DTranspose,paddle.nn.layer.conv.Conv3DTranspose
**Convlution3D transpose layer** **Convlution3D transpose layer**
...@@ -664,52 +685,30 @@ class Conv3DTranspose(layers.Layer): ...@@ -664,52 +685,30 @@ class Conv3DTranspose(layers.Layer):
Example: Example:
- Input:
Input shape: :math:`(N, C_{in}, D_{in}, H_{in}, W_{in})`
Filter shape: :math:`(C_{in}, C_{out}, D_f, H_f, W_f)`
- Output:
Output shape: :math:`(N, C_{out}, D_{out}, H_{out}, W_{out})`
Where
.. math::
D^\prime_{out} &= (D_{in} - 1) * strides[0] - 2 * paddings[0] + dilations[0] * (D_f - 1) + 1 \\\\
H^\prime_{out} &= (H_{in} - 1) * strides[1] - 2 * paddings[1] + dilations[1] * (H_f - 1) + 1 \\\\
W^\prime_{out} &= (W_{in} - 1) * strides[2] - 2 * paddings[2] + dilations[2] * (W_f - 1) + 1 \\\\
D_{out} &\in [ D^\prime_{out}, D^\prime_{out} + strides[0] ] \\\\
H_{out} &\in [ H^\prime_{out}, H^\prime_{out} + strides[1] ] \\\\
**Note**: **Note**:
The conv3d_transpose can be seen as the backward of the conv3d. For conv3d, The conv_transpose3d can be seen as the backward of the conv3d. For conv3d,
when stride > 1, conv3d maps multiple input shape to the same output shape, when stride > 1, conv3d maps multiple input shape to the same output shape,
so for conv3d_transpose, when stride > 1, input shape maps multiple output shape. so for conv_transpose3d, when stride > 1, input shape maps multiple output shape.
If output_size is None, :math:`H_{out} = H^\prime_{out}, :math:`H_{out} = \ If output_size is None, :math:`H_{out} = H^\prime_{out}, :math:`H_{out} = \
H^\prime_{out}, W_{out} = W^\prime_{out}`; else, the :math:`D_{out}` of the output H^\prime_{out}, W_{out} = W^\prime_{out}`; else, the :math:`D_{out}` of the output
size must between :math:`D^\prime_{out}` and :math:`D^\prime_{out} + strides[0]`, size must between :math:`D^\prime_{out}` and :math:`D^\prime_{out} + strides[0]`,
the :math:`H_{out}` of the output size must between :math:`H^\prime_{out}` the :math:`H_{out}` of the output size must between :math:`H^\prime_{out}`
and :math:`H^\prime_{out} + strides[1]`, and the :math:`W_{out}` of the output size must and :math:`H^\prime_{out} + strides[1]`, and the :math:`W_{out}` of the output size must
between :math:`W^\prime_{out}` and :math:`W^\prime_{out} + strides[2]`, between :math:`W^\prime_{out}` and :math:`W^\prime_{out} + strides[2]`,
conv3d_transpose can compute the kernel size automatically. conv_transpose3d can compute the kernel size automatically.
Parameters: Parameters:
num_channels(int): The number of channels in the input image. in_channels(int): The number of channels in the input image.
num_filters(int): The number of the filter. It is as same as the output out_channels(int): The number of channels produced by the convolution.
image channel. kernel_size(int|list|tuple): The kernel size. If kernel_size is a tuple,
filter_size(int|tuple): The filter size. If filter_size is a tuple, it must contain three integers, (kernel_size_D, kernel_size_H, kernel_size_W).
it must contain three integers, (filter_size_D, filter_size_H, filter_size_W). Otherwise, the kernel will be a square.
Otherwise, the filter will be a square. stride(int|list|tuple, optional): The stride size. It means the stride in transposed convolution.
output_size(int or tuple, optional): The output image size. If output size is a If stride is a tuple, it must contain three integers, (stride_depth, stride_height,
tuple, it must contain two integers, (image_H, image_W). None if use stride_width). Otherwise, stride_depth = stride_height = stride_width = stride.
filter_size, padding, and stride to calculate output_size. The default value is 1.
if output_size and filter_size are specified at the same time, They
should follow the formula above. Default: None.
padding(int|str|tuple|list, optional): The padding size. Padding coule be in one of the following forms. padding(int|str|tuple|list, optional): The padding size. Padding coule be in one of the following forms.
1. a string in ['valid', 'same']. 1. a string in ['valid', 'same'].
2. an int, which means each spartial dimension(depth, height, width) is zero paded by size of `padding` 2. an int, which means each spartial dimension(depth, height, width) is zero paded by size of `padding`
...@@ -717,11 +716,9 @@ class Conv3DTranspose(layers.Layer): ...@@ -717,11 +716,9 @@ class Conv3DTranspose(layers.Layer):
4. a list[int] or tuple[int] whose length is 2 * number of spartial dimensions. It has the form [pad_before, pad_after, pad_before, pad_after, ...] for all spartial dimensions. 4. a list[int] or tuple[int] whose length is 2 * number of spartial dimensions. It has the form [pad_before, pad_after, pad_before, pad_after, ...] for all spartial dimensions.
5. a list or tuple of pairs of ints. It has the form [[pad_before, pad_after], [pad_before, pad_after], ...]. Note that, the batch dimension and channel dimension are also included. Each pair of integers correspond to the amount of padding for a dimension of the input. Padding in batch dimension and channel dimension should be [0, 0] or (0, 0). 5. a list or tuple of pairs of ints. It has the form [[pad_before, pad_after], [pad_before, pad_after], ...]. Note that, the batch dimension and channel dimension are also included. Each pair of integers correspond to the amount of padding for a dimension of the input. Padding in batch dimension and channel dimension should be [0, 0] or (0, 0).
The default value is 0. The default value is 0.
stride(int|tuple, optional): The stride size. It means the stride in transposed convolution. output_padding(int|list|tuple, optional): Additional size added to one side
If stride is a tuple, it must contain three integers, (stride_depth, stride_height, of each dimension in the output shape. Default: 0.
stride_width). Otherwise, stride_depth = stride_height = stride_width = stride. dilation(int|list|tuple, optional): The dilation size. If dilation is a tuple, it must
The default value is 1.
dilation(int|tuple, optional): The dilation size. If dilation is a tuple, it must
contain three integers, (dilation_D, dilation_H, dilation_W). Otherwise, the contain three integers, (dilation_D, dilation_H, dilation_W). Otherwise, the
dilation_D = dilation_H = dilation_W = dilation. The default value is 1. dilation_D = dilation_H = dilation_W = dilation. The default value is 1.
groups(int, optional): The groups number of the Conv3d transpose layer. Inspired by groups(int, optional): The groups number of the Conv3d transpose layer. Inspired by
...@@ -730,7 +727,7 @@ class Conv3DTranspose(layers.Layer): ...@@ -730,7 +727,7 @@ class Conv3DTranspose(layers.Layer):
first half of the input channels, while the second half of the first half of the input channels, while the second half of the
filters is only connected to the second half of the input channels. filters is only connected to the second half of the input channels.
The default value is 1. The default value is 1.
param_attr (ParamAttr, optional): The parameter attribute for learnable parameters/weights weight_attr (ParamAttr, optional): The parameter attribute for learnable parameters/weights
of conv3d_transpose. If it is set to None or one attribute of ParamAttr, conv3d_transpose of conv3d_transpose. If it is set to None or one attribute of ParamAttr, conv3d_transpose
will create ParamAttr as param_attr. If the Initializer of the param_attr will create ParamAttr as param_attr. If the Initializer of the param_attr
is not set, the parameter is initialized with Xavier. The default value is None. is not set, the parameter is initialized with Xavier. The default value is None.
...@@ -739,10 +736,11 @@ class Conv3DTranspose(layers.Layer): ...@@ -739,10 +736,11 @@ class Conv3DTranspose(layers.Layer):
If it is set to None or one attribute of ParamAttr, conv3d_transpose If it is set to None or one attribute of ParamAttr, conv3d_transpose
will create ParamAttr as bias_attr. If the Initializer of the bias_attr will create ParamAttr as bias_attr. If the Initializer of the bias_attr
is not set, the bias is initialized zero. The default value is None. is not set, the bias is initialized zero. The default value is None.
use_cudnn(bool, optional): Use cudnn kernel or not, it is valid only when the cudnn output_size(int|list|tuple, optional): The output image size. If output size is a
library is installed. The default value is True. tuple, it must contain two integers, (image_H, image_W). None if use
act (str, optional): Activation type, if it is set to None, activation is not appended. filter_size, padding, and stride to calculate output_size.
The default value is None. if output_size and filter_size are specified at the same time, They
should follow the formula above. Default: None.
data_format (str, optional): Data format that specifies the layout of input. data_format (str, optional): Data format that specifies the layout of input.
It can be "NCDHW" or "NDHWC". Default: "NCDHW". It can be "NCDHW" or "NDHWC". Default: "NCDHW".
...@@ -751,8 +749,19 @@ class Conv3DTranspose(layers.Layer): ...@@ -751,8 +749,19 @@ class Conv3DTranspose(layers.Layer):
**bias** (Parameter): the learnable bias of this layer. **bias** (Parameter): the learnable bias of this layer.
Returns: Shape:
None. - x: :math:`(N, C_{in}, D_{in}, H_{in}, W_{in})`
- output: :math:`(N, C_{out}, D_{out}, H_{out}, W_{out})`
Where
.. math::
D^\prime_{out} &= (D_{in} - 1) * strides[0] - 2 * paddings[0] + dilations[0] * (kernel_size[0] - 1) + 1 \\\\
H^\prime_{out} &= (H_{in} - 1) * strides[1] - 2 * paddings[1] + dilations[1] * (kernel_size[1] - 1) + 1 \\\\
W^\prime_{out} &= (W_{in} - 1) * strides[2] - 2 * paddings[2] + dilations[2] * (kernel_size[2] - 1) + 1 \\\\
Raises: Raises:
ValueError: If the shapes of input, filter_size, stride, padding and ValueError: If the shapes of input, filter_size, stride, padding and
...@@ -762,14 +771,14 @@ class Conv3DTranspose(layers.Layer): ...@@ -762,14 +771,14 @@ class Conv3DTranspose(layers.Layer):
.. code-block:: python .. code-block:: python
import numpy as np import numpy as np
from paddle import fluid import paddle
import paddle.fluid.dygraph as dg import paddle.nn as nn
from paddle import nn
x = np.random.uniform(-1, 1, (2, 4, 8, 8, 8)).astype('float32') x = np.random.uniform(-1, 1, (2, 4, 8, 8, 8)).astype('float32')
place = fluid.CPUPlace()
with dg.guard(place): paddle.disable_static()
x_var = dg.to_variable(x)
x_var = paddle.to_tensor(x)
conv = nn.Conv3DTranspose(4, 6, (3, 3, 3)) conv = nn.Conv3DTranspose(4, 6, (3, 3, 3))
y_var = conv(x_var) y_var = conv(x_var)
y_np = y_var.numpy() y_np = y_var.numpy()
...@@ -779,69 +788,47 @@ class Conv3DTranspose(layers.Layer): ...@@ -779,69 +788,47 @@ class Conv3DTranspose(layers.Layer):
""" """
def __init__(self, def __init__(self,
num_channels, in_channels,
num_filters, out_channels,
filter_size, kernel_size,
output_size=None,
padding=0,
stride=1, stride=1,
padding=0,
output_padding=0,
dilation=1, dilation=1,
groups=1, groups=1,
param_attr=None, weight_attr=None,
bias_attr=None, bias_attr=None,
use_cudnn=True, data_format="NCDHW"):
act=None, super(ConvTranspose3d, self).__init__(
data_format="NCDHW", in_channels,
dtype='float32'): out_channels,
super(Conv3DTranspose, self).__init__() kernel_size,
if not isinstance(use_cudnn, bool): True,
raise ValueError("use_cudnn should be True or False") 3,
assert param_attr is not False, "param_attr should not be False in conv3d_transpose." stride=stride,
self._num_channels = num_channels padding=padding,
self._num_filters = num_filters dilation=dilation,
self._groups = groups output_padding=output_padding,
self._use_cudnn = use_cudnn groups=groups,
self._act = act weight_attr=weight_attr,
self._dtype = dtype bias_attr=bias_attr,
self._data_format = data_format data_format=data_format)
self._stride = utils.convert_to_list(stride, 3, 'stride') def forward(self, x, output_size):
self._dilation = utils.convert_to_list(dilation, 3, 'dilation')
self._filter_size = utils.convert_to_list(filter_size, 3, 'filter_size')
channel_last = (data_format == "NDHWC")
self._padding = padding
if output_size is None: if output_size is None:
self._output_size = output_size output_padding = self.output_padding
elif isinstance(output_size, (list, tuple, int)):
self._output_size = utils.convert_to_list(output_size, 3,
'output_size')
else: else:
raise ValueError( output_padding = 0
"output_size should be int, ot list[int] or tuple[int]")
self._param_attr = param_attr out = F.conv_transpose3d(
self._bias_attr = bias_attr x,
filter_shape = [num_channels, num_filters // groups] + self._filter_size
self.weight = self.create_parameter(
dtype=self._dtype, shape=filter_shape, attr=self._param_attr)
self.bias = self.create_parameter(
attr=self._bias_attr,
shape=[self._num_filters],
dtype=self._dtype,
is_bias=True)
def forward(self, input):
out = F.conv3d_transpose(
input,
self.weight, self.weight,
bias=self.bias, bias=self.bias,
output_size=self._output_size,
padding=self._padding, padding=self._padding,
output_padding=output_padding,
stride=self._stride, stride=self._stride,
dilation=self._dilation, dilation=self._dilation,
groups=self._groups, groups=self._groups,
use_cudnn=self._use_cudnn, output_size=output_size,
act=self._act,
data_format=self._data_format) data_format=self._data_format)
return out return out
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或