未验证 提交 fc5c432a 编写于 作者: T Tao Luo 提交者: GitHub

Merge pull request #6787 from luotao1/seq_pool_doc

add python wrapper for sequence_pool
......@@ -170,6 +170,18 @@ sequence_pool
:noindex:
sequence_first_step
-------------------
.. autofunction:: paddle.v2.fluid.layers.sequence_first_step
:noindex:
sequence_last_step
------------------
.. autofunction:: paddle.v2.fluid.layers.sequence_last_step
:noindex:
pool2d
------
.. autofunction:: paddle.v2.fluid.layers.pool2d
......
......@@ -13,7 +13,8 @@ __all__ = [
'crf_decoding', 'cos_sim', 'cross_entropy', 'square_error_cost', 'accuracy',
'chunk_eval', 'sequence_conv', 'conv2d', 'sequence_pool', 'pool2d',
'batch_norm', 'beam_search_decode', 'conv2d_transpose', 'sequence_expand',
'lstm_unit', 'reduce_sum', 'reduce_mean'
'lstm_unit', 'reduce_sum', 'reduce_mean', 'sequence_first_step',
'sequence_last_step'
]
......@@ -574,9 +575,53 @@ def conv2d(input,
def sequence_pool(input, pool_type, **kwargs):
"""
This function add the operator for sequence pooling.
This is applied on top of the input using pool_type mentioned
in the parameters.
This function add the operator for sequence pooling.
It pools features of all time-steps of each instance, and is applied
on top of the input using pool_type mentioned in the parameters.
It supports four pool_type:
- average: :math:`Out[i] = \\frac{\sum_i X_i}{N}`
- sum: :math:`Out[i] = \sum_jX_{ij}`
- sqrt: :math:`Out[i] = \\frac{\sum_jX_{ij}}{\sqrt{len(X_i)}}`
- max: :math:`Out[i] = max(X_i)`
.. code-block:: text
x is a 1-level LoDTensor:
x.lod = [[0, 2, 5, 7]]
x.data = [1, 3, 2, 4, 6, 5, 1]
x.dims = [7, 1]
then output is a Tensor:
out.dim = [3, 1]
with condition len(x.lod[-1]) - 1 == out.dims[0]
for different pool_type:
average: out.data = [2, 4, 3], where 2=(1+3)/2, 4=(2+4+6)/3, 3=(5+1)/2
sum : out.data = [4, 12, 6], where 4=1+3, 12=2+4+6, 6=5+1
sqrt : out.data = [2.82, 6.93, 4.24], where 2.82=(1+3)/sqrt(2),
6.93=(2+4+6)/sqrt(3), 4.24=(5+1)/sqrt(2)
max : out.data = [3, 6, 5], where 3=max(1,3), 6=max(2,4,6), 5=max(5,1)
Args:
input(variable): The input variable which is a LoDTensor.
pool_type (string): The pooling type of sequence_pool.
It supports average, sum, sqrt and max.
Returns:
The sequence pooling variable which is a Tensor.
Examples:
.. code-block:: python
x = fluid.layers.data(name='x', shape=[7, 1],
dtype='float32', lod_level=1)
avg_x = fluid.layers.sequence_pool(input=x, pool_type='average')
sum_x = fluid.layers.sequence_pool(input=x, pool_type='sum')
sqrt_x = fluid.layers.sequence_pool(input=x, pool_type='sqrt')
max_x = fluid.layers.sequence_pool(input=x, pool_type='max')
"""
helper = LayerHelper('sequence_pool', input=input, **kwargs)
dtype = helper.input_dtype()
......@@ -593,6 +638,72 @@ def sequence_pool(input, pool_type, **kwargs):
return pool_out
def sequence_first_step(input, **kwargs):
"""
This funciton get the first step of sequence.
.. code-block:: text
x is a 1-level LoDTensor:
x.lod = [[0, 2, 5, 7]]
x.data = [1, 3, 2, 4, 6, 5, 1]
x.dims = [7, 1]
then output is a Tensor:
out.dim = [3, 1]
with condition len(x.lod[-1]) - 1 == out.dims[0]
out.data = [1, 2, 5], where 1=first(1,3), 2=first(2,4,6), 5=first(5,1)
Args:
input(variable): The input variable which is a LoDTensor.
Returns:
The sequence's first step variable which is a Tensor.
Examples:
.. code-block:: python
x = fluid.layers.data(name='x', shape=[7, 1],
dtype='float32', lod_level=1)
x_first_step = fluid.layers.sequence_first_step(input=x)
"""
return sequence_pool(input=input, pool_type="first")
def sequence_last_step(input, **kwargs):
"""
This funciton get the last step of sequence.
.. code-block:: text
x is a 1-level LoDTensor:
x.lod = [[0, 2, 5, 7]]
x.data = [1, 3, 2, 4, 6, 5, 1]
x.dims = [7, 1]
then output is a Tensor:
out.dim = [3, 1]
with condition len(x.lod[-1]) - 1 == out.dims[0]
out.data = [3, 6, 1], where 3=last(1,3), 6=last(2,4,6), 1=last(5,1)
Args:
input(variable): The input variable which is a LoDTensor.
Returns:
The sequence's last step variable which is a Tensor.
Examples:
.. code-block:: python
x = fluid.layers.data(name='x', shape=[7, 1],
dtype='float32', lod_level=1)
x_last_step = fluid.layers.sequence_last_step(input=x)
"""
return sequence_pool(input=input, pool_type="last")
def pool2d(input,
pool_size,
pool_type,
......
......@@ -33,7 +33,7 @@ def encoder_decoder():
fc1 = fluid.layers.fc(input=src_embedding, size=hidden_dim * 4, act='tanh')
lstm_hidden0, lstm_0 = layers.dynamic_lstm(input=fc1, size=hidden_dim * 4)
encoder_out = layers.sequence_pool(input=lstm_hidden0, pool_type="last")
encoder_out = layers.sequence_last_step(input=lstm_hidden0)
# decoder
trg_language_word = layers.data(
......
......@@ -63,8 +63,7 @@ class TestDynRNN(unittest.TestCase):
all_timesteps = fluid.layers.array_to_lod_tensor(
x=out, table=rank_table)
last = fluid.layers.sequence_pool(
input=all_timesteps, pool_type='last')
last = fluid.layers.sequence_last_step(input=all_timesteps)
logits = fluid.layers.fc(input=last, size=1, act=None)
loss = fluid.layers.sigmoid_cross_entropy_with_logits(
x=logits, label=label)
......@@ -101,7 +100,7 @@ class TestDynRNN(unittest.TestCase):
rnn.update_memory(mem, out_)
rnn.output(out_)
last = fluid.layers.sequence_pool(input=rnn(), pool_type='last')
last = fluid.layers.sequence_last_step(input=rnn())
logits = fluid.layers.fc(input=last, size=1, act=None)
label = fluid.layers.data(name='label', shape=[1], dtype='float32')
loss = fluid.layers.sigmoid_cross_entropy_with_logits(
......
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