提交 3b5e6d9c 编写于 作者: J jinyuKing

update text.py

上级 3e180974
...@@ -1898,14 +1898,16 @@ class SequenceTagging(fluid.dygraph.Layer): ...@@ -1898,14 +1898,16 @@ class SequenceTagging(fluid.dygraph.Layer):
crf_decode = self.crf_decoding(input=emission, length=lengths) crf_decode = self.crf_decoding(input=emission, length=lengths)
return crf_decode, lengths return crf_decode, lengths
class Conv1dPoolLayer(Layer): class Conv1dPoolLayer(Layer):
""" """
This interface is used to construct a callable object of the ``Conv1DPoolLayer`` class.The ``Conv1DPoolLayer`` is composed of a ``Conv2D`` and a ``Pool2D`` . This interface is used to construct a callable object of the ``Conv1DPoolLayer`` class.
For more details, refer to code examples.The ``Conv1DPoolLayer`` layer calculates the output based on the input, filter and strides, paddings, dilations, The ``Conv1DPoolLayer`` class does a ``Conv1D`` and a ``Pool1D`` .For more details,
groups,global_pooling, pool_type,ceil_mode,exclusive parameters.Input and Output are in NCH format, where N is batch size, C is the number of the feature map, refer to code examples.The ``Conv1DPoolLayer`` layer calculates the output based on the
H is the height of the feature map.The data type of Input data and Output data is 'float32' or 'float64'. input, filter and strides, paddings, dilations, groups,global_pooling, pool_type,ceil_mode,
exclusive parameters.
Args: Parameters:
input(Variable):3-D Tensor, shape is [N, C, H], data type can be float32 or float64 input(Variable):3-D Tensor, shape is [N, C, H], data type can be float32 or float64
num_channels(int): The number of channels in the input data. num_channels(int): The number of channels in the input data.
num_filters(int): The number of filters. It is the same as the output channels. num_filters(int): The number of filters. It is the same as the output channels.
...@@ -1915,147 +1917,174 @@ class Conv1dPoolLayer(Layer): ...@@ -1915,147 +1917,174 @@ class Conv1dPoolLayer(Layer):
pool_stride (int): The stride size of the pool layer in Conv1DPoolLayer. Default: 1 pool_stride (int): The stride size of the pool layer in Conv1DPoolLayer. Default: 1
conv_padding (int): The padding size of the conv Layer in Conv1DPoolLayer. Default: 0 conv_padding (int): The padding size of the conv Layer in Conv1DPoolLayer. Default: 0
pool_padding (int): The padding of pool layer in Conv1DPoolLayer. Default: 0 pool_padding (int): The padding of pool layer in Conv1DPoolLayer. Default: 0
pool_type (str): Pooling type can be `max` for max-pooling or `avg` for average-pooling. Default: math:`max` pool_type (str): Pooling type can be `max` for max-pooling or `avg` for average-pooling.
global_pooling (bool): Whether to use the global pooling. If global_pooling = true, pool_size and pool_padding while be ignored. Default: False Default: math:`max`
global_pooling (bool): Whether to use the global pooling. If global_pooling = true,
pool_size and pool_padding while be ignored. Default: False
dilation (int): The dilation size of the conv Layer. Default: 1. dilation (int): The dilation size of the conv Layer. Default: 1.
groups (int): The groups number of the conv Layer. According to grouped convolution in Alex Krizhevsky's Deep CNN paper: when group=2, groups (int): The groups number of the conv Layer. According to grouped convolution in
the first half of the filters is only connected to the first half of the input channels, while the second half of the filters is only Alex Krizhevsky's Deep CNN paper: when group=2, the first half of the filters is
connected to the second half of the input channels. Default: 1. only connected to the first half of the input channels, while the second half
param_attr (ParamAttr|None): The parameter attribute for learnable parameters/weights of conv layer. If it is set to None or one attribute of of the filters is only connected to the second half of the input channels.
ParamAttr, conv2d will create ParamAttr as param_attr. If the Initializer of the param_attr is not set, the parameter is initialized Default: 1.
with :`Normal(0.0, std)`,and the :`std` is :`(\\frac{2.0 }{filter\_elem\_num})^{0.5}`.Default: None. param_attr (ParamAttr|None): The parameter attribute for learnable parameters/weights
bias_attr (ParamAttr|bool|None): The parameter attribute for the bias of conv.If it is set to False, no bias will be added to the output units. of conv layer. If it is set to None or one attribute of ParamAttr, conv2d will
If it is set to None or one attribute of ParamAttr, conv2d will create ParamAttr as bias_attr. If the Initializer of the bias_attr is not create ParamAttr as param_attr. If the Initializer of the param_attr
set, the bias is initialized zero. Default: None. is not set, the parameter is initialized with :`Normal(0.0, std)`,and
name(str, optional): The default value is None. Normally there is no need for user to set this property. Default: None the :`std` is :`(\\frac{2.0 }{filter\_elem\_num})^{0.5}`.Default: None.
act (str): Activation type for conv layer, if it is set to None, activation is not appended. Default: None. bias_attr (ParamAttr|bool|None): The parameter attribute for the bias of conv.If it
use_cudnn (bool): Use cudnn kernel or not, it is valid only when the cudnn library is installed. Default: False is set to False, no bias will be added to the output units.If it is set to
ceil_mode (bool, optional): Whether to use the ceil function to calculate output height and width. None or one attribute of ParamAttr, conv2d will create ParamAttr as bias_attr.
False is the default. If it is set to False, the floor function will be used. Default: False. If the Initializer of the bias_attr is not set, the bias is initialized zero.
exclusive (bool, optional): Whether to exclude padding points in average pooling mode. Default: True. Default: None.
act (str): Activation type for conv layer, if it is set to None, activation is not appended.
Return: Default: None.
3-D Tensor, the result of input after conv and pool, with the same data type as :`input` use_cudnn (bool): Use cudnn kernel or not, it is valid only when the cudnn library
is installed. Default: False
Return Type: ceil_mode (bool, optional): Whether to use the ceil function to calculate output
Variable height and width.False is the default. If it is set to False, the floor function
will be used. Default: False.
exclusive (bool, optional): Whether to exclude padding points in average pooling mode.
Default: True.
Example: Example:
```python
import paddle.fluid as fluid .. code-block:: python
from hapi.text import Conv1dPoolLayer
import paddle.fluid as fluid
test=np.random.uniform(-1,1,[2,3,4]).astype('float32') from paddle.incubate.hapi.text import Conv1dPoolLayer
with fluid.dygraph.guard():
paddle_input=to_variable(test) test=np.random.uniform(-1,1,[2,3,4]).astype('float32')
print(paddle_input.shape) with fluid.dygraph.guard():
cov2d=Conv1dPoolLayer(3,4,2,2) paddle_input=to_variable(test)
paddle_out=cov2d(paddle_input) cov2d=Conv1dPoolLayer(3,4,2,2)
print(paddle_out.shape)#[2,4,2] paddle_out=cov2d(paddle_input)
print(paddle_out.shape)#[2,4,2]
```
""" """
def __init__(self,
num_channels, def __init__(
num_filters, self,
filter_size, num_channels,
pool_size, num_filters,
conv_stride=1, filter_size,
pool_stride=1, pool_size,
conv_padding=0, conv_stride=1,
pool_padding=0, pool_stride=1,
pool_type='max', conv_padding=0,
global_pooling=False, pool_padding=0,
dilation=1, pool_type='max',
groups=None, global_pooling=False,
param_attr=None, dilation=1,
bias_attr=None, groups=None,
act=None, param_attr=None,
use_cudnn=False, bias_attr=None,
ceil_mode=False, act=None,
exclusive=True, use_cudnn=False,
): ceil_mode=False,
exclusive=True, ):
super(Conv1dPoolLayer, self).__init__() super(Conv1dPoolLayer, self).__init__()
self._conv2d = Conv2D(num_channels=num_channels, self._conv2d = Conv2D(
num_filters=num_filters, num_channels=num_channels,
filter_size=[filter_size,1], num_filters=num_filters,
stride=[conv_stride,1], filter_size=[filter_size, 1],
padding=[conv_padding,0], stride=[conv_stride, 1],
dilation=[dilation,1], padding=[conv_padding, 0],
groups=groups, dilation=[dilation, 1],
param_attr=param_attr, groups=groups,
bias_attr=bias_attr, param_attr=param_attr,
use_cudnn=use_cudnn, bias_attr=bias_attr,
act=act) use_cudnn=use_cudnn,
self._pool2d = Pool2D(pool_size=[pool_size,1], act=act)
pool_type=pool_type, self._pool2d = Pool2D(
pool_stride=[pool_stride,1], pool_size=[pool_size, 1],
pool_padding=[pool_padding,0], pool_type=pool_type,
global_pooling=global_pooling, pool_stride=[pool_stride, 1],
use_cudnn=use_cudnn, pool_padding=[pool_padding, 0],
ceil_mode=ceil_mode, global_pooling=global_pooling,
exclusive=exclusive use_cudnn=use_cudnn,
) ceil_mode=ceil_mode,
exclusive=exclusive)
def forward(self, inputs): def forward(self, inputs):
x = fluid.layers.unsqueeze(inputs,axes=[-1]) """
Performs :Code:`Conv1dPoolLayer.forward` receives input data. After a conv and a pool,
the result will be output.
Parameters:
inputs(Variable): Inputs are 3-D Tensor, shape is [N, C, H] , where N is batch size,
C is the number of the feature map, H is the height of the feature map.The data
type of Input data is 'float32' or 'float64'.
Returns:
3-D Tensor, with the same data type as :`input`
"""
x = fluid.layers.unsqueeze(inputs, axes=[-1])
x = self._conv2d(x) x = self._conv2d(x)
x = self._pool2d(x) x = self._pool2d(x)
x = fluid.layers.squeeze(x, axes=[-1]) x = fluid.layers.squeeze(x, axes=[-1])
return x return x
class CNNEncoder(Layer): class CNNEncoder(Layer):
""" """
This interface is used to construct a callable object of the ``CNNEncoder`` class.The ``CNNEncoder`` is composed of a ``Embedding`` and a ``Conv1dPoolLayer`` . This interface is used to construct a callable object of the ``CNNEncoder`` class.The ``CNNEncoder``
For more details, refer to code examples. The ``CNNEncoder`` layer calculates the output based on the input, dict_size and emb_dim, filter_size, num_filters, is composed of ``Conv1dPoolLayer`` .``CNNEncoder`` can define every Conv1dPoolLayer with different
use_cuda, is_sparse, param_attr parameters. The type of Input data is a 3-D Tensor .The data type of Input data is 'float32'. Output data are in NCH or same parameters. The ``Conv1dPoolLayer`` in ``CNNEncoder`` is parallel.The result of every
format, where N is batch size, C is the number of the feature map, H is the height of the feature map. The data type of Output data is 'float32' or 'float64'. ``Conv1dPoolLayer`` will concat as the final output.For more details, refer to code examples.
The ``CNNEncoder`` layer calculates the output based on the input, dict_size and emb_dim,
Args: filter_size, num_filters, use_cuda, is_sparse, param_attr parameters.
num_channels(int|list|tuple): The number of channels in the input data.If num_channels is a list or tuple, the length of num_channels must equal layer_num.If num_channels
is a int, all conv1dpoollayer's num_channels are the value of num_channels. Parameters:
num_filters(int|list|tuple): The number of filters. It is the same as the output channels. If num_filters is a list or tuple, the length of num_filters must equal layer_num.If num_filters num_channels(int|list|tuple): The number of channels in the input data.If num_channels is
is a int, all conv1dpoollayer's num_filters are the value of num_filters. a list or tuple, the length of num_channels must equal layer_num.If num_channels
filter_size(int|list|tuple): The filter size of Conv1DPoolLayer in CNNEncoder. If filter_size is a list or tuple, the length of filter_size must equal layer_num.If filter_size is a int, all conv1dpoollayer's num_channels are the value of num_channels.
is a int, all conv1dpoollayer's filter_size are the value of filter_size. num_filters(int|list|tuple): The number of filters. It is the same as the output channels.
pool_size(int|list|tuple): The pooling size of Conv1DPoolLayer in CNNEncoder.If pool_size is a list or tuple, the length of pool_size must equal layer_num.If pool_size If num_filters is a list or tuple, the length of num_filters must equal layer_num.
is a int, all conv1dpoollayer's pool_size are the value of pool_size. If num_filters is a int, all conv1dpoollayer's num_filters are the value of num_filters.
layer_num(int): The number of conv1dpoolLayer used in CNNEncoder. filter_size(int|list|tuple): The filter size of Conv1DPoolLayer in CNNEncoder. If filter_size
conv_stride(int|list|tuple): The stride size of the conv Layer in Conv1DPoolLayer. If conv_stride is a list or tuple, the length of conv_stride must equal layer_num.If conv_stride is a list or tuple, the length of filter_size must equal layer_num.If filter_size is a
is a int, all conv1dpoollayer's conv_stride are the value of conv_stride. Default: 1 int, all conv1dpoollayer's filter_size are the value of filter_size.
pool_stride(int|list|tuple): The stride size of the pool layer in Conv1DPoolLayer. If pool_stride is a list or tuple, the length of pool_stride must equal layer_num.If pool_stride pool_size(int|list|tuple): The pooling size of Conv1DPoolLayer in CNNEncoder.If pool_size is
is a int, all conv1dpoollayer's pool_stride are the value of pool_stride. Default: 1 a list or tuple, the length of pool_size must equal layer_num.If pool_size is a int,
conv_padding(int|list|tuple): The padding size of the conv Layer in Conv1DPoolLayer.If conv_padding is a list or tuple, the length of conv_padding must equal layer_num.If conv_padding all conv1dpoollayer's pool_size are the value of pool_size.
is a int, all conv1dpoollayer's conv_padding are the value of conv_padding. Default: 0 layer_num(int): The number of conv1dpoolLayer used in CNNEncoder.
pool_padding(int|list|tuple): The padding of pool layer in Conv1DPoolLayer. If pool_padding is a list or tuple, the length of pool_padding must equal layer_num.If pool_padding conv_stride(int|list|tuple): The stride size of the conv Layer in Conv1DPoolLayer. If
is a int, all conv1dpoollayer's pool_padding are the value of pool_padding. Default: 0 conv_stride is a list or tuple, the length of conv_stride must equal layer_num.
use_cudnn (bool): Use cudnn kernel or not, it is valid only when the cudnn library is installed. Default: False If conv_stride is a int, all conv1dpoollayer's conv_stride are the value of
act (str|list|tuple): Activation type for `Conv1dPoollayer` layer, if it is set to None, activation is not appended. Default: None. conv_stride. Default: 1
pool_stride(int|list|tuple): The stride size of the pool layer in Conv1DPoolLayer. If
Return: pool_stride is a list or tuple, the length of pool_stride must equal layer_num.
3-D Tensor, the result of input after embedding and conv1dPoollayer If pool_stride is a int, all conv1dpoollayer's pool_stride are the value of
pool_stride. Default: 1
Return Type: conv_padding(int|list|tuple): The padding size of the conv Layer in Conv1DPoolLayer.
Variable If conv_padding is a list or tuple, the length of conv_padding must equal layer_num.
If conv_padding is a int, all conv1dpoollayer's conv_padding are the value of
conv_padding. Default: 0
pool_padding(int|list|tuple): The padding of pool layer in Conv1DPoolLayer. If pool_padding is
a list or tuple, the length of pool_padding must equal layer_num.If pool_padding is a
int, all conv1dpoollayer's pool_padding are the value of pool_padding. Default: 0
use_cudnn (bool): Use cudnn kernel or not, it is valid only when the cudnn library is installed.
Default: False
act (str|list|tuple): Activation type for `Conv1dPoollayer` layer, if it is set to None,
activation is not appended. Default: None.
Example: Example:
```python
import paddle.fluid as fluid
from hapi.text import CNNEncoder
test=np.random.uniform(1,5,[2,3,4]).astype('float32') .. code-block:: python
with fluid.dygraph.guard():
paddle_input=to_variable(test) import paddle.fluid as fluid
#print(paddle_input.shape) from paddle.incubate.hapi.text import Conv1dPoolLayer
cov2d=CNNEncoder(3,4,2,2,3)
paddle_out=cov2d(paddle_input)
print(paddle_out)#[2,12,2]
``` test=np.random.uniform(-1,1,[2,3,4]).astype('float32')
with fluid.dygraph.guard():
paddle_input=to_variable(test)
cov2d=CNNEncoder(3,4,2,2,3)
paddle_out=cov2d(paddle_input)
print(paddle_out)#[2,12,2]
""" """
def __init__(self, def __init__(self,
num_channels, num_channels,
num_filters, num_filters,
...@@ -2067,33 +2096,55 @@ class CNNEncoder(Layer): ...@@ -2067,33 +2096,55 @@ class CNNEncoder(Layer):
conv_padding=0, conv_padding=0,
pool_padding=0, pool_padding=0,
use_cudnn=False, use_cudnn=False,
act=None act=None):
):
super(CNNEncoder, self).__init__() super(CNNEncoder, self).__init__()
self.num_channels=num_channels self.num_channels = num_channels
self.num_filters=num_filters self.num_filters = num_filters
self.filter_size=filter_size self.filter_size = filter_size
self.pool_size=pool_size self.pool_size = pool_size
self.layer_num=layer_num self.layer_num = layer_num
self.conv_stride=conv_stride self.conv_stride = conv_stride
self.pool_stride=pool_stride self.pool_stride = pool_stride
self.conv_padding=conv_padding self.conv_padding = conv_padding
self.pool_padding=pool_padding self.pool_padding = pool_padding
self.use_cudnn=use_cudnn self.use_cudnn = use_cudnn
self.act=act self.act = act
self.conv_layer = fluid.dygraph.LayerList([Conv1dPoolLayer(num_channels=self.num_channels if isinstance(self.num_channels,int) else self.num_channels[i], self.conv_layer = fluid.dygraph.LayerList([
num_filters=self.num_filters if isinstance(self.num_channels,int) else self.num_filters [i], Conv1dPoolLayer(
filter_size=self.filter_size if isinstance(self.filter_size,int) else self.filter_size[i], num_channels=self.num_channels if
pool_size=self.pool_size if isinstance(self.pool_size,int) else self.pool_size[i], isinstance(self.num_channels, int) else self.num_channels[i],
conv_stride=self.conv_stride if isinstance(self.conv_stride,int) else self.conv_stride[i], num_filters=self.num_filters
pool_stride=self.pool_stride if isinstance(self.pool_stride,int) else self.pool_stride[i], if isinstance(self.num_channels, int) else self.num_filters[i],
conv_padding= self.conv_padding if isinstance(self.conv_padding,int) else self.conv_padding[i], filter_size=self.filter_size
pool_padding=self.pool_padding if isinstance(self.pool_padding,int) else self.pool_padding[i], if isinstance(self.filter_size, int) else self.filter_size[i],
act=self.act[i] if isinstance(self.act,(list,tuple)) else self.act, pool_size=self.pool_size
use_cudnn=self.use_cudnn if isinstance(self.pool_size, int) else self.pool_size[i],
) for i in range(layer_num)]) conv_stride=self.conv_stride
if isinstance(self.conv_stride, int) else self.conv_stride[i],
pool_stride=self.pool_stride
if isinstance(self.pool_stride, int) else self.pool_stride[i],
conv_padding=self.conv_padding if
isinstance(self.conv_padding, int) else self.conv_padding[i],
pool_padding=self.pool_padding if
isinstance(self.pool_padding, int) else self.pool_padding[i],
act=self.act[i]
if isinstance(self.act, (list, tuple)) else self.act,
use_cudnn=self.use_cudnn) for i in range(layer_num)
])
def forward(self, input): def forward(self, input):
res=[Conv1dPoolLayer(input) for Conv1dPoolLayer in self.conv_layer] """
out=fluid.layers.concat(input=res,axis=1) Performs :Code:`CNNEncoder.forward` receives input data.The result of every ``Conv1dPoolLayer``
return out will store in a list. Then, the results will be concated as the final output.
\ No newline at end of file
Parameters:
inputs(Variable): Inputs are 3-D Tensor, shape is [N, C, H] , where N is batch size,
C is the number of the feature map, H is the height of the feature map.The data
type of Input data is 'float32' or 'float64'.
Returns:
3-D Tensor, with the same data type as :`input`
"""
res = [Conv1dPoolLayer(input) for Conv1dPoolLayer in self.conv_layer]
out = fluid.layers.concat(input=res, axis=1)
return out
Markdown is supported
0% .
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