nn.py 175.8 KB
Newer Older
1
#   Copyright (c) 2018 PaddlePaddle Authors. All Rights Reserved.
D
dzhwinter 已提交
2
#
D
dzhwinter 已提交
3 4 5
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
D
dzhwinter 已提交
6
#
D
dzhwinter 已提交
7
#     http://www.apache.org/licenses/LICENSE-2.0
D
dzhwinter 已提交
8
#
D
dzhwinter 已提交
9 10 11 12 13
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
Y
Yu Yang 已提交
14
"""
15
All layers just related to the neural network. 
Y
Yu Yang 已提交
16 17 18 19 20
"""

from ..layer_helper import LayerHelper
from ..initializer import Normal, Constant
from ..framework import Variable
Y
yangyaming 已提交
21
from ..param_attr import ParamAttr
Y
yuyang18 已提交
22
from layer_function_generator import autodoc, templatedoc
Y
yangyaming 已提交
23
from tensor import concat
C
chengduoZH 已提交
24
import utils
Y
yuyang18 已提交
25
import random
Y
Yu Yang 已提交
26 27

__all__ = [
Y
ying 已提交
28 29 30
    'fc',
    'embedding',
    'dynamic_lstm',
Y
Yibing Liu 已提交
31
    'dynamic_lstmp',
G
guosheng 已提交
32
    'dynamic_gru',
Y
ying 已提交
33 34 35 36 37 38 39 40 41
    'gru_unit',
    'linear_chain_crf',
    'crf_decoding',
    'cos_sim',
    'cross_entropy',
    'square_error_cost',
    'chunk_eval',
    'sequence_conv',
    'conv2d',
42
    'conv3d',
Y
ying 已提交
43
    'sequence_pool',
44 45
    'sequence_softmax',
    'softmax',
Y
ying 已提交
46
    'pool2d',
47
    'pool3d',
Y
ying 已提交
48 49 50
    'batch_norm',
    'beam_search_decode',
    'conv2d_transpose',
51
    'conv3d_transpose',
Y
ying 已提交
52 53 54 55 56 57
    'sequence_expand',
    'lstm_unit',
    'reduce_sum',
    'reduce_mean',
    'reduce_max',
    'reduce_min',
58
    'reduce_prod',
Y
ying 已提交
59 60 61 62
    'sequence_first_step',
    'sequence_last_step',
    'dropout',
    'split',
63 64
    'ctc_greedy_decoder',
    'edit_distance',
Y
ying 已提交
65 66
    'l2_normalize',
    'matmul',
Q
qingqing01 已提交
67
    'topk',
Y
ying 已提交
68 69
    'warpctc',
    'sequence_reshape',
70
    'transpose',
71
    'im2sequence',
72
    'nce',
Q
Qiao Longfei 已提交
73
    'beam_search',
74
    'row_conv',
75
    'multiplex',
G
guosheng 已提交
76
    'layer_norm',
77 78
    'softmax_with_cross_entropy',
    'smooth_l1',
79
    'one_hot',
Y
Yu Yang 已提交
80
    'autoincreased_step_counter',
C
caoying03 已提交
81
    'reshape',
Y
yangyaming 已提交
82
    'lod_reset',
D
dragonwarrior 已提交
83
    'lrn',
G
guosheng 已提交
84
    'pad',
85
    'label_smooth',
86
    'roi_pool',
W
whs 已提交
87
    'dice_loss',
F
fengjiayi 已提交
88 89
    'image_resize',
    'image_resize_short',
B
baiyf 已提交
90
    'resize_bilinear',
W
whs 已提交
91
    'gather',
92
    'random_crop',
93
    'mean_iou',
Y
Yu Yang 已提交
94 95 96 97 98 99 100 101
]


def fc(input,
       size,
       num_flatten_dims=1,
       param_attr=None,
       bias_attr=None,
102
       use_mkldnn=False,
Y
Yu Yang 已提交
103
       act=None,
J
Jacek Czaja 已提交
104
       is_test=False,
105
       name=None):
Y
Yu Yang 已提交
106
    """
107
    **Fully Connected Layer**
Y
Yu Yang 已提交
108

C
caoying03 已提交
109
    The fully connected layer can take multiple tensors as its inputs. It
R
ranqiu 已提交
110 111 112 113 114 115
    creates a variable called weights for each input tensor, which represents
    a fully connected weight matrix from each input unit to each output unit.
    The fully connected layer multiplies each input tensor with its coresponding
    weight to produce an output Tensor. If multiple input tensors are given,
    the results of multiple multiplications will be sumed up. If bias_attr is
    not None, a bias variable will be created and added to the output. Finally,
Y
ying 已提交
116
    if activation is not None, it will be applied to the output as well.
C
caoying03 已提交
117

C
caoying03 已提交
118
    This process can be formulated as follows:
119 120 121

    .. math::

122
        Out = Act({\sum_{i=0}^{N-1}X_iW_i + b})
123 124 125

    In the above equation:

C
caoying03 已提交
126 127 128 129
    * :math:`N`: Number of the input.
    * :math:`X_i`: The input tensor.
    * :math:`W`: The weights created by this layer.
    * :math:`b`: The bias parameter created by this layer (if needed).
130
    * :math:`Act`: The activation function.
C
caoying03 已提交
131
    * :math:`Out`: The output tensor.
Y
Yu Yang 已提交
132 133

    Args:
R
ranqiu 已提交
134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150
        input (Variable|list of Variable): The input tensor(s) of this layer, and the dimension of
            the input tensor(s) is at least 2.
        size(int): The number of output units in this layer.
        num_flatten_dims (int, default 1): The fc layer can accept an input tensor with more than
            two dimensions. If this happens, the multidimensional tensor will first be flattened
            into a 2-dimensional matrix. The parameter `num_flatten_dims` determines how the input
            tensor is flattened: the first `num_flatten_dims` (inclusive, index starts from 1)
            dimensions will be flatten to form the first dimension of the final matrix (height of
            the matrix), and the rest `rank(X) - num_flatten_dims` dimensions are flattened to
            form the second dimension of the final matrix (width of the matrix). For example, suppose
            `X` is a 6-dimensional tensor with a shape [2, 3, 4, 5, 6], and `num_flatten_dims` = 3.
            Then, the flattened matrix will have a shape [2 x 3 x 4, 5 x 6] = [24, 30].
        param_attr (ParamAttr|list of ParamAttr, default None): The parameter attribute for learnable
            parameters/weights of this layer.
        bias_attr (ParamAttr|list of ParamAttr, default None): The parameter attribute for the bias
            of this layer. If it is set to None, no bias will be added to the output units.
        act (str, default None): Activation to be applied to the output of this layer.
J
Jacek Czaja 已提交
151
        is_test(bool): A flag indicating whether execution is in test phase.
M
mozga-intel 已提交
152 153
        use_mkldnn(bool): Use mkldnn kernel or not, it is valid only when the mkldnn
            library is installed. Default: False
R
ranqiu 已提交
154
        name (str, default None): The name of this layer.
Y
Yu Yang 已提交
155

156
    Returns:
R
ranqiu 已提交
157
        A tensor variable storing the transformation result.
158 159

    Raises:
C
caoying03 已提交
160
        ValueError: If rank of the input tensor is less than 2.
161 162 163 164

    Examples:
        .. code-block:: python

F
stash  
fengjiayi 已提交
165 166
          data = fluid.layers.data(
              name="data", shape=[32, 32], dtype="float32")
167
          fc = fluid.layers.fc(input=data, size=1000, act="tanh")
Y
Yu Yang 已提交
168
    """
C
caoying03 已提交
169

C
caoying03 已提交
170
    helper = LayerHelper("fc", **locals())
Y
Yu Yang 已提交
171 172 173 174

    dtype = helper.input_dtype()

    mul_results = []
175 176
    for input_var, param_attr in helper.iter_inputs_and_params():
        input_shape = input_var.shape
Y
Yu Yang 已提交
177 178 179
        param_shape = [
            reduce(lambda a, b: a * b, input_shape[num_flatten_dims:], 1)
        ] + [size]
Y
ying 已提交
180

Y
Yu Yang 已提交
181
        w = helper.create_parameter(
182 183
            attr=param_attr, shape=param_shape, dtype=dtype, is_bias=False)
        tmp = helper.create_tmp_variable(dtype)
184
        helper.append_op(
185 186 187
            type="mul",
            inputs={"X": input_var,
                    "Y": w},
188
            outputs={"Out": tmp},
M
mozga-intel 已提交
189 190
            attrs={"x_num_col_dims": num_flatten_dims,
                   "y_num_col_dims": 1})
191 192 193 194
        mul_results.append(tmp)

    if len(mul_results) == 1:
        pre_bias = mul_results[0]
195
    else:
196 197 198 199 200 201 202
        pre_bias = helper.create_tmp_variable(dtype)
        helper.append_op(
            type="sum", inputs={"X": mul_results}, outputs={"Out": pre_bias})
    # add bias
    pre_activation = helper.append_bias_op(pre_bias, dim_start=num_flatten_dims)
    # add activation
    return helper.append_activation(pre_activation)
Y
Yu Yang 已提交
203 204


205 206 207
def embedding(input,
              size,
              is_sparse=False,
208
              is_distributed=False,
209 210 211
              padding_idx=None,
              param_attr=None,
              dtype='float32'):
Y
Yu Yang 已提交
212
    """
213 214
    **Embedding Layer**

215
    This layer is used to lookup embeddings of IDs, provided by :attr:`input`, in
216 217
    a lookup table. The result of this lookup is the embedding of each ID in the
    :attr:`input`.
218 219 220

    All the input variables are passed in as local variables to the LayerHelper
    constructor.
Y
Yu Yang 已提交
221 222

    Args:
223 224 225 226 227
        input(Variable): The tensor variable containing the IDs.
        size(tuple|list): The shape of the look up table parameter. It should
            have two elements which indicate the size of the dictionary of
            embeddings and the size of each embedding vector respectively.
        is_sparse(bool): The flag indicating whether to use sparse update.
228
        is_distributed (bool): Whether to run lookup table from remote parameter server.
229 230
        padding_idx(int|long|None): If :attr:`None`, it makes no effect to lookup.
            Otherwise the given :attr:`padding_idx` indicates padding the output
231 232
            with zeros whenever lookup encounters it in :attr:`input`. If
            :math:`padding_idx < 0`, the padding_idx to use in lookup is
233 234
            :math:`size[0] + dim`.
        param_attr(ParamAttr): Parameters for this layer
235
        dtype(np.dtype|core.VarDesc.VarType|str): The type of data : float32, float_16, int etc
Y
Yu Yang 已提交
236

237 238 239
    Returns:
        Variable: The tensor variable storing the embeddings of the \
                  supplied inputs.
Y
Yu Yang 已提交
240

241 242
    Examples:
        .. code-block:: python
Y
Yu Yang 已提交
243

C
chengduoZH 已提交
244
          dict_size = len(dataset.ids)
245
          data = fluid.layers.data(name='ids', shape=[32, 32], dtype='float32')
C
chengduoZH 已提交
246
          fc = fluid.layers.embedding(input=data, size=[dict_size, 16])
Y
Yu Yang 已提交
247 248 249 250 251 252
    """

    helper = LayerHelper('embedding', **locals())
    w = helper.create_parameter(
        attr=helper.param_attr, shape=size, dtype=dtype, is_bias=False)
    tmp = helper.create_tmp_variable(dtype)
253 254
    padding_idx = -1 if padding_idx is None else padding_idx if padding_idx >= 0 else (
        size[0] + padding_idx)
Y
Yu Yang 已提交
255 256 257 258 259
    helper.append_op(
        type='lookup_table',
        inputs={'Ids': input,
                'W': w},
        outputs={'Out': tmp},
260 261 262 263 264
        attrs={
            'is_sparse': is_sparse,
            'is_distributed': is_distributed,
            'padding_idx': padding_idx
        })
Y
Yu Yang 已提交
265 266 267 268 269
    return tmp


def dynamic_lstm(input,
                 size,
Y
Yancey 已提交
270 271
                 h_0=None,
                 c_0=None,
Y
Yu Yang 已提交
272 273 274 275 276 277 278
                 param_attr=None,
                 bias_attr=None,
                 use_peepholes=True,
                 is_reverse=False,
                 gate_activation='sigmoid',
                 cell_activation='tanh',
                 candidate_activation='tanh',
279 280
                 dtype='float32',
                 name=None):
Y
Yibing Liu 已提交
281 282 283 284 285 286
    """
    **Dynamic LSTM Layer**

    The defalut implementation is diagonal/peephole connection
    (https://arxiv.org/pdf/1402.1128.pdf), the formula is as follows:

Y
Yibing Liu 已提交
287
    .. math::
Y
Yibing Liu 已提交
288

289
        i_t & = \sigma(W_{ix}x_{t} + W_{ih}h_{t-1} + W_{ic}c_{t-1} + b_i)
Y
Yibing Liu 已提交
290

291
        f_t & = \sigma(W_{fx}x_{t} + W_{fh}h_{t-1} + W_{fc}c_{t-1} + b_f)
Y
Yibing Liu 已提交
292

293
        \\tilde{c_t} & = act_g(W_{cx}x_t + W_{ch}h_{t-1} + b_c)
Y
Yibing Liu 已提交
294

295 296 297
        o_t & = \sigma(W_{ox}x_{t} + W_{oh}h_{t-1} + W_{oc}c_t + b_o)

        c_t & = f_t \odot c_{t-1} + i_t \odot \\tilde{c_t}
Y
Yibing Liu 已提交
298

Y
Yibing Liu 已提交
299
        h_t & = o_t \odot act_h(c_t)
Y
Yibing Liu 已提交
300

301
    where the :math:`W` terms denote weight matrices (e.g. :math:`W_{xi}` is
302
    the matrix of weights from the input gate to the input), :math:`W_{ic}, \
303 304 305
    W_{fc}, W_{oc}` are diagonal weight matrices for peephole connections. In
    our implementation, we use vectors to reprenset these diagonal weight
    matrices. The :math:`b` terms denote bias vectors (:math:`b_i` is the input
Y
Yibing Liu 已提交
306
    gate bias vector), :math:`\sigma` is the non-linear activations, such as
307 308
    logistic sigmoid function, and :math:`i, f, o` and :math:`c` are the input
    gate, forget gate, output gate, and cell activation vectors, respectively,
309 310
    all of which have the same size as the cell output activation vector :math:`h`.

311 312 313 314
    The :math:`\odot` is the element-wise product of the vectors. :math:`act_g`
    and :math:`act_h` are the cell input and cell output activation functions
    and `tanh` is usually used for them. :math:`\\tilde{c_t}` is also called
    candidate hidden state, which is computed based on the current input and
315 316 317
    the previous hidden state.

    Set `use_peepholes` to `False` to disable peephole connection. The formula
Y
Yibing Liu 已提交
318 319 320
    is omitted here, please refer to the paper
    http://www.bioinf.jku.at/publications/older/2604.pdf for details.

Y
Yibing Liu 已提交
321 322 323
    Note that these :math:`W_{xi}x_{t}, W_{xf}x_{t}, W_{xc}x_{t}, W_{xo}x_{t}`
    operations on the input :math:`x_{t}` are NOT included in this operator.
    Users can choose to use fully-connect layer before LSTM layer.
Y
Yibing Liu 已提交
324 325

    Args:
326 327 328 329
        input(Variable): The input of dynamic_lstm layer, which supports
                         variable-time length input sequence. The underlying
                         tensor in this Variable is a matrix with shape
                         (T X 4D), where T is the total time steps in this
Y
Yibing Liu 已提交
330 331
                         mini-batch, D is the hidden size.
        size(int): 4 * hidden size.
Y
Yancey 已提交
332 333 334 335 336 337 338
        h_0(Variable): The initial hidden state is an optional input, default is zero.
                       This is a tensor with shape (N x D), where N is the
                       batch size and D is the hidden size.
        c_0(Variable): The initial cell state is an optional input, default is zero.
                       This is a tensor with shape (N x D), where N is the
                       batch size. `h_0` and `c_0` can be NULL but only at the same time.

339
        param_attr(ParamAttr|None): The parameter attribute for the learnable
340
                               hidden-hidden weights.
Y
Yibing Liu 已提交
341 342 343

                               - Weights = {:math:`W_{ch}, W_{ih}, \
                                                W_{fh}, W_{oh}`}
344 345 346
                               - The shape is (D x 4D), where D is the hidden
                                 size.
        bias_attr(ParamAttr|None): The bias attribute for the learnable bias
347 348 349
                              weights, which contains two parts, input-hidden
                              bias weights and peephole connections weights if
                              setting `use_peepholes` to `True`.
Y
Yibing Liu 已提交
350

351
                              1. `use_peepholes = False`
Y
Yibing Liu 已提交
352
                                - Biases = {:math:`b_c, b_i, b_f, b_o`}.
353
                                - The shape is (1 x 4D).
354
                              2. `use_peepholes = True`
Y
Yibing Liu 已提交
355 356
                                - Biases = { :math:`b_c, b_i, b_f, b_o, W_{ic}, \
                                                 W_{fc}, W_{oc}`}.
357
                                - The shape is (1 x 7D).
358
        use_peepholes(bool): Whether to enable diagonal/peephole connections,
Y
Yibing Liu 已提交
359 360
                             default `True`.
        is_reverse(bool): Whether to compute reversed LSTM, default `False`.
361 362
        gate_activation(str): The activation for input gate, forget gate and
                              output gate. Choices = ["sigmoid", "tanh", "relu",
Y
Yibing Liu 已提交
363
                              "identity"], default "sigmoid".
364
        cell_activation(str): The activation for cell output. Choices = ["sigmoid",
Y
Yibing Liu 已提交
365 366
                              "tanh", "relu", "identity"], default "tanh".
        candidate_activation(str): The activation for candidate hidden state.
F
stash  
fengjiayi 已提交
367 368
                              Choices = ["sigmoid", "tanh",
                                  "relu", "identity"],
Y
Yibing Liu 已提交
369 370
                              default "tanh".
        dtype(str): Data type. Choices = ["float32", "float64"], default "float32".
371 372
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.
Y
Yibing Liu 已提交
373 374

    Returns:
Y
Yibing Liu 已提交
375 376
        tuple: The hidden state, and cell state of LSTM. The shape of both \
        is (T x D), and lod is the same with the `input`.
Y
Yibing Liu 已提交
377

Y
Yibing Liu 已提交
378
    Examples:
Y
Yibing Liu 已提交
379 380
        .. code-block:: python

Y
Yibing Liu 已提交
381 382
            hidden_dim = 512
            forward_proj = fluid.layers.fc(input=input_seq, size=hidden_dim * 4,
383
                                           act=None, bias_attr=None)
Y
Yibing Liu 已提交
384 385
            forward, _ = fluid.layers.dynamic_lstm(
                input=forward_proj, size=hidden_dim * 4, use_peepholes=False)
Y
Yibing Liu 已提交
386
    """
387

Y
Yu Yang 已提交
388 389 390 391 392 393 394 395 396 397 398 399 400 401
    helper = LayerHelper('lstm', **locals())
    size = size / 4
    weight = helper.create_parameter(
        attr=helper.param_attr, shape=[size, 4 * size], dtype=dtype)
    bias_size = [1, 7 * size]
    if not use_peepholes:
        bias_size[1] = 4 * size
    bias = helper.create_parameter(
        attr=helper.bias_attr, shape=bias_size, dtype=dtype, is_bias=True)

    hidden = helper.create_tmp_variable(dtype)
    cell = helper.create_tmp_variable(dtype)
    batch_gate = helper.create_tmp_variable(dtype)
    batch_cell_pre_act = helper.create_tmp_variable(dtype)
Y
Yancey 已提交
402 403 404 405 406 407 408 409 410 411
    inputs = {'Input': input, 'Weight': weight, 'Bias': bias}
    batch_size = input.shape[0]
    if h_0:
        assert h_0.shape == (batch_size, size), \
            'The shape of h0 should be (batch_size, %d)' % size
        inputs['H0'] = h_0
    if c_0:
        assert c_0.shape == (batch_size, size), \
            'The shape of c0 should be (batch_size, %d)' % size
        inputs['C0'] = c_0
Y
Yu Yang 已提交
412 413 414

    helper.append_op(
        type='lstm',
Y
Yancey 已提交
415
        inputs=inputs,
Y
Yu Yang 已提交
416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431
        outputs={
            'Hidden': hidden,
            'Cell': cell,
            'BatchGate': batch_gate,
            'BatchCellPreAct': batch_cell_pre_act
        },
        attrs={
            'use_peepholes': use_peepholes,
            'is_reverse': is_reverse,
            'gate_activation': gate_activation,
            'cell_activation': cell_activation,
            'candidate_activation': candidate_activation
        })
    return hidden, cell


Y
Yibing Liu 已提交
432 433 434 435 436 437 438 439 440 441 442
def dynamic_lstmp(input,
                  size,
                  proj_size,
                  param_attr=None,
                  bias_attr=None,
                  use_peepholes=True,
                  is_reverse=False,
                  gate_activation='sigmoid',
                  cell_activation='tanh',
                  candidate_activation='tanh',
                  proj_activation='tanh',
443 444
                  dtype='float32',
                  name=None):
Y
Yibing Liu 已提交
445 446 447
    """
    **Dynamic LSTMP Layer**

448 449 450 451 452 453
    LSTMP (LSTM with recurrent projection) layer has a separate projection
    layer after the LSTM layer, projecting the original hidden state to a
    lower-dimensional one, which is proposed to reduce the number of total
    parameters and furthermore computational complexity for the LSTM,
    espeacially for the case that the size of output units is relative
    large (https://research.google.com/pubs/archive/43905.pdf).
Y
Yibing Liu 已提交
454 455 456 457 458

    The formula is as follows:

    .. math::

459
        i_t & = \sigma(W_{ix}x_{t} + W_{ir}r_{t-1} + W_{ic}c_{t-1} + b_i)
Y
Yibing Liu 已提交
460

461
        f_t & = \sigma(W_{fx}x_{t} + W_{fr}r_{t-1} + W_{fc}c_{t-1} + b_f)
Y
Yibing Liu 已提交
462

463
        \\tilde{c_t} & = act_g(W_{cx}x_t + W_{cr}r_{t-1} + b_c)
Y
Yibing Liu 已提交
464

465
        o_t & = \sigma(W_{ox}x_{t} + W_{or}r_{t-1} + W_{oc}c_t + b_o)
Y
Yibing Liu 已提交
466

467
        c_t & = f_t \odot c_{t-1} + i_t \odot \\tilde{c_t}
Y
Yibing Liu 已提交
468

469
        h_t & = o_t \odot act_h(c_t)
Y
Yibing Liu 已提交
470

471
        r_t & = \overline{act_h}(W_{rh}h_t)
Y
Yibing Liu 已提交
472

Y
Yibing Liu 已提交
473 474 475 476 477 478
    In the above formula:

    * :math:`W`: Denotes weight matrices (e.g. :math:`W_{xi}` is \
          the matrix of weights from the input gate to the input).
    * :math:`W_{ic}`, :math:`W_{fc}`, :math:`W_{oc}`: Diagonal weight \
          matrices for peephole connections. In our implementation, \
479
          we use vectors to reprenset these diagonal weight matrices.
Y
Yibing Liu 已提交
480
    * :math:`b`: Denotes bias vectors (e.g. :math:`b_i` is the input gate \
481
          bias vector).
Y
Yibing Liu 已提交
482 483 484
    * :math:`\sigma`: The activation, such as logistic sigmoid function.
    * :math:`i, f, o` and :math:`c`: The input gate, forget gate, output \
          gate, and cell activation vectors, respectively, all of which have \
485
          the same size as the cell output activation vector :math:`h`.
Y
Yibing Liu 已提交
486
    * :math:`h`: The hidden state.
487
    * :math:`r`: The recurrent projection of the hidden state.
Y
Yibing Liu 已提交
488 489
    * :math:`\\tilde{c_t}`: The candidate hidden state, whose \
          computation is based on the current input and previous hidden state.
490
    * :math:`\odot`: The element-wise product of the vectors.
Y
Yibing Liu 已提交
491
    * :math:`act_g` and :math:`act_h`: The cell input and cell output \
492
          activation functions and `tanh` is usually used for them.
Y
Yibing Liu 已提交
493 494
    * :math:`\overline{act_h}`: The activation function for the projection \
          output, usually using `identity` or same as :math:`act_h`.
Y
Yibing Liu 已提交
495 496 497 498

    Set `use_peepholes` to `False` to disable peephole connection. The formula
    is omitted here, please refer to the paper
    http://www.bioinf.jku.at/publications/older/2604.pdf for details.
499

Y
Yibing Liu 已提交
500 501 502 503 504 505 506 507 508 509 510 511
    Note that these :math:`W_{xi}x_{t}, W_{xf}x_{t}, W_{xc}x_{t}, W_{xo}x_{t}`
    operations on the input :math:`x_{t}` are NOT included in this operator.
    Users can choose to use fully-connected layer before LSTMP layer.

    Args:
        input(Variable): The input of dynamic_lstmp layer, which supports
                         variable-time length input sequence. The underlying
                         tensor in this Variable is a matrix with shape
                         (T X 4D), where T is the total time steps in this
                         mini-batch, D is the hidden size.
        size(int): 4 * hidden size.
        proj_size(int): The size of projection output.
512
        param_attr(ParamAttr|None): The parameter attribute for the learnable
Y
Yibing Liu 已提交
513 514
                               hidden-hidden weight and projection weight.

515 516
                               - Hidden-hidden weight = {:math:`W_{ch}, W_{ih}, \
                                                W_{fh}, W_{oh}`}.
517 518
                               - The shape of hidden-hidden weight is (P x 4D),
                                 where P is the projection size and D the hidden
Y
Yibing Liu 已提交
519 520
                                 size.
                               - Projection weight = {:math:`W_{rh}`}.
521 522
                               - The shape of projection weight is (D x P).
        bias_attr(ParamAttr|None): The bias attribute for the learnable bias
Y
Yibing Liu 已提交
523 524 525 526 527 528
                              weights, which contains two parts, input-hidden
                              bias weights and peephole connections weights if
                              setting `use_peepholes` to `True`.

                              1. `use_peepholes = False`
                                - Biases = {:math:`b_c, b_i, b_f, b_o`}.
529
                                - The shape is (1 x 4D).
Y
Yibing Liu 已提交
530 531 532
                              2. `use_peepholes = True`
                                - Biases = { :math:`b_c, b_i, b_f, b_o, W_{ic}, \
                                                 W_{fc}, W_{oc}`}.
533
                                - The shape is (1 x 7D).
Y
Yibing Liu 已提交
534 535 536 537 538 539 540 541 542
        use_peepholes(bool): Whether to enable diagonal/peephole connections,
                             default `True`.
        is_reverse(bool): Whether to compute reversed LSTM, default `False`.
        gate_activation(str): The activation for input gate, forget gate and
                              output gate. Choices = ["sigmoid", "tanh", "relu",
                              "identity"], default "sigmoid".
        cell_activation(str): The activation for cell output. Choices = ["sigmoid",
                              "tanh", "relu", "identity"], default "tanh".
        candidate_activation(str): The activation for candidate hidden state.
F
stash  
fengjiayi 已提交
543 544
                              Choices = ["sigmoid", "tanh",
                                  "relu", "identity"],
Y
Yibing Liu 已提交
545 546
                              default "tanh".
        proj_activation(str): The activation for projection output.
F
stash  
fengjiayi 已提交
547 548
                              Choices = ["sigmoid", "tanh",
                                  "relu", "identity"],
Y
Yibing Liu 已提交
549 550
                              default "tanh".
        dtype(str): Data type. Choices = ["float32", "float64"], default "float32".
551 552
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.
Y
Yibing Liu 已提交
553 554

    Returns:
555 556
        tuple: The projection of hidden state, and cell state of LSTMP. The \
               shape of projection is (T x P), for the cell state which is \
Y
Yibing Liu 已提交
557 558 559 560 561
               (T x D), and both LoD is the same with the `input`.

    Examples:
        .. code-block:: python

Y
Yibing Liu 已提交
562
            hidden_dim, proj_dim = 512, 256
Y
Yibing Liu 已提交
563 564
            fc_out = fluid.layers.fc(input=input_seq, size=hidden_dim * 4,
                                     act=None, bias_attr=None)
565 566 567
            proj_out, _ = fluid.layers.dynamic_lstmp(input=fc_out,
                                                     size=hidden_dim * 4,
                                                     proj_size=proj_dim,
Y
Yibing Liu 已提交
568 569 570 571
                                                     use_peepholes=False,
                                                     is_reverse=True,
                                                     cell_activation="tanh",
                                                     proj_activation="tanh")
Y
Yibing Liu 已提交
572
    """
573

Y
Yibing Liu 已提交
574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619
    helper = LayerHelper('lstmp', **locals())
    size = size / 4
    weight = helper.create_parameter(
        attr=helper.param_attr, shape=[proj_size, 4 * size], dtype=dtype)
    proj_weight = helper.create_parameter(
        attr=helper.param_attr, shape=[size, proj_size], dtype=dtype)
    bias_size = [1, 7 * size]
    if not use_peepholes:
        bias_size[1] = 4 * size
    bias = helper.create_parameter(
        attr=helper.bias_attr, shape=bias_size, dtype=dtype, is_bias=True)

    projection = helper.create_tmp_variable(dtype)
    cell = helper.create_tmp_variable(dtype)
    ordered_proj0 = helper.create_tmp_variable(dtype)
    batch_hidden = helper.create_tmp_variable(dtype)
    batch_gate = helper.create_tmp_variable(dtype)
    batch_cell_pre_act = helper.create_tmp_variable(dtype)

    helper.append_op(
        type='lstmp',
        inputs={
            'Input': input,
            'Weight': weight,
            'ProjWeight': proj_weight,
            'Bias': bias
        },
        outputs={
            'Projection': projection,
            'Cell': cell,
            'OrderedP0': ordered_proj0,
            'BatchHidden': batch_hidden,
            'BatchGate': batch_gate,
            'BatchCellPreAct': batch_cell_pre_act
        },
        attrs={
            'use_peepholes': use_peepholes,
            'is_reverse': is_reverse,
            'gate_activation': gate_activation,
            'cell_activation': cell_activation,
            'candidate_activation': candidate_activation,
            'proj_activation': proj_activation
        })
    return projection, cell


G
guosheng 已提交
620 621 622 623 624 625 626 627 628 629 630
def dynamic_gru(input,
                size,
                param_attr=None,
                bias_attr=None,
                is_reverse=False,
                gate_activation='sigmoid',
                candidate_activation='tanh',
                h_0=None):
    """
    **Dynamic GRU Layer**

631
    Refer to `Empirical Evaluation of Gated Recurrent Neural Networks on
G
guosheng 已提交
632
    Sequence Modeling <https://arxiv.org/abs/1412.3555>`_
633

G
guosheng 已提交
634 635 636 637 638 639 640 641 642
    The formula is as follows:

    .. math::

        u_t & = act_g(W_{ux}x_{t} + W_{uh}h_{t-1} + b_u)

        r_t & = act_g(W_{rx}x_{t} + W_{rh}h_{t-1} + b_r)

        \\tilde{h_t} & = act_c(W_{cx}x_{t} + W_{ch}(r_t \odot h_{t-1}) + b_c)
643

G
guosheng 已提交
644
        h_t & = (1-u_t) \odot h_{t-1} + u_t \odot \\tilde{h_t}
645

G
guosheng 已提交
646
    The :math:`\odot` is the element-wise product of the vectors. :math:`act_g`
647 648
    is the update gate and reset gate activation function and :math:`sigmoid`
    is usually used for it. :math:`act_c` is the activation function for
G
guosheng 已提交
649 650 651 652
    candidate hidden state and :math:`tanh` is usually used for it.

    Note that these :math:`W_{ux}x_{t}, W_{rx}x_{t}, W_{cx}x_{t}` operations on
    the input :math:`x_{t}` are NOT included in this operator. Users can choose
653
    to use fully-connect layer before GRU layer.
G
guosheng 已提交
654 655

    Args:
656 657
        input(Variable): The input of dynamic_gru layer, which supports
            variable-time length input sequence. The underlying tensor in this
G
guosheng 已提交
658
            Variable is a matrix with shape :math:`(T \\times 3D)`, where
659
            :math:`T` is the total time steps in this mini-batch, :math:`D`
G
guosheng 已提交
660 661
            is the hidden size.
        size(int): The dimension of the gru cell.
662
        param_attr(ParamAttr|None): The parameter attribute for the learnable
G
guosheng 已提交
663 664
            hidden-hidden weight matrix. Note:

665
            - The shape of the weight matrix is :math:`(T \\times 3D)`, where
G
guosheng 已提交
666
              :math:`D` is the hidden size.
667
            - All elements in the weight matrix can be divided into two parts.
G
guosheng 已提交
668
              The first part are weights of the update gate and reset gate with
669
              shape :math:`(D \\times 2D)`, and the second part are weights for
G
guosheng 已提交
670
              candidate hidden state with shape :math:`(D \\times D)`.
671
        bias_attr(ParamAttr): The parameter attribute for learnable the
G
guosheng 已提交
672
            hidden-hidden bias.
673
        is_reverse(bool): Whether to compute reversed GRU, default
G
guosheng 已提交
674 675 676
            :attr:`False`.
        gate_activation(str): The activation for update gate and reset gate.
            Choices = ["sigmoid", "tanh", "relu", "identity"], default "sigmoid".
677
        candidate_activation(str): The activation for candidate hidden state.
G
guosheng 已提交
678
            Choices = ["sigmoid", "tanh", "relu", "identity"], default "tanh".
679
        h_0 (Variable): The hidden output of the first time step.
G
guosheng 已提交
680 681

    Returns:
G
guosheng 已提交
682 683
        Variable: The hidden state of GRU. The shape is :math:`(T \\times D)`, \
            and lod is the same with the input.
684

G
guosheng 已提交
685 686 687 688 689 690 691 692 693 694 695 696 697 698 699
    Examples:
        .. code-block:: python

            hidden_dim = 512
            x = fluid.layers.fc(input=data, size=hidden_dim * 3)
            hidden = fluid.layers.dynamic_gru(input=x, dim=hidden_dim)
    """

    helper = LayerHelper('gru', **locals())
    dtype = helper.input_dtype()

    weight = helper.create_parameter(
        attr=helper.param_attr, shape=[size, 3 * size], dtype=dtype)
    bias = helper.create_parameter(
        attr=helper.bias_attr, shape=[1, 3 * size], dtype=dtype, is_bias=True)
Y
Yancey 已提交
700
    batch_size = input.shape[0]
G
guosheng 已提交
701 702 703
    inputs = {'Input': input, 'Weight': weight, 'Bias': bias}
    if h_0 != None:
        assert h_0.shape == (
Y
Yancey 已提交
704 705 706
            batch_size, size
        ), 'The shape of h0 should be(batch_size, %d)' % size
        inputs['H0'] = h_0
G
guosheng 已提交
707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729

    hidden = helper.create_tmp_variable(dtype)
    batch_gate = helper.create_tmp_variable(dtype)
    batch_reset_hidden_prev = helper.create_tmp_variable(dtype)
    batch_hidden = helper.create_tmp_variable(dtype)

    helper.append_op(
        type='gru',
        inputs=inputs,
        outputs={
            'Hidden': hidden,
            'BatchGate': batch_gate,
            'BatchResetHiddenPrev': batch_reset_hidden_prev,
            'BatchHidden': batch_hidden
        },
        attrs={
            'is_reverse': is_reverse,
            'gate_activation': gate_activation,
            'activation': candidate_activation
        })
    return hidden


Y
Yu Yang 已提交
730 731 732
def gru_unit(input,
             hidden,
             size,
733 734
             param_attr=None,
             bias_attr=None,
Y
Yu Yang 已提交
735
             activation='tanh',
736
             gate_activation='sigmoid'):
Y
Yu Yang 已提交
737
    """
738
    GRU unit layer. The equation of a gru step is:
Y
Yu Yang 已提交
739

740 741
        .. math::
            u_t & = actGate(xu_{t} + W_u h_{t-1} + b_u)
Y
Yu Yang 已提交
742

743
            r_t & = actGate(xr_{t} + W_r h_{t-1} + b_r)
Y
Yu Yang 已提交
744

745
            m_t & = actNode(xm_t + W_c dot(r_t, h_{t-1}) + b_m)
746

747
            h_t & = dot((1-u_t), m_t) + dot(u_t, h_{t-1})
748 749

    The inputs of gru unit includes :math:`z_t`, :math:`h_{t-1}`. In terms
750 751 752
    of the equation above, the :math:`z_t` is split into 3 parts -
    :math:`xu_t`, :math:`xr_t` and :math:`xm_t`. This means that in order to
    implement a full GRU unit operator for an input, a fully
753 754
    connected layer has to be applied, such that :math:`z_t = W_{fc}x_t`.

755 756
    The terms :math:`u_t` and :math:`r_t` represent the update and reset gates
    of the GRU cell. Unlike LSTM, GRU has one lesser gate. However, there is
757 758 759
    an intermediate candidate hidden output, which is denoted by :math:`m_t`.
    This layer has three outputs :math:`h_t`, :math:`dot(r_t, h_{t-1})`
    and concatenation of :math:`u_t`, :math:`r_t` and :math:`m_t`.
760 761 762 763 764

    Args:
        input (Variable): The fc transformed input value of current step.
        hidden (Variable): The hidden value of lstm unit from previous step.
        size (integer): The input dimension value.
765 766
        param_attr (ParamAttr): The weight parameters for gru unit. Default: None
        bias_attr (ParamAttr): The bias parameters for gru unit. Default: None
767 768 769 770
        activation (string): The activation type for cell (actNode).
                             Default: 'tanh'
        gate_activation (string): The activation type for gates (actGate).
                                  Default: 'sigmoid'
Y
Yu Yang 已提交
771

772 773 774 775 776 777
    Returns:
        tuple: The hidden value, reset-hidden value and gate values.

    Examples:

        .. code-block:: python
Y
Yu Yang 已提交
778

779
             # assuming we have x_t_data and prev_hidden of size=10
780
             x_t = fluid.layers.fc(input=x_t_data, size=30)
781 782
             hidden_val, r_h_val, gate_val = fluid.layers.gru_unit(input=x_t,
                                                    hidden = prev_hidden)
Y
Yu Yang 已提交
783 784 785 786 787 788 789 790 791 792 793 794 795 796 797

    """
    activation_dict = dict(
        identity=0,
        sigmoid=1,
        tanh=2,
        relu=3, )
    activation = activation_dict[activation]
    gate_activation = activation_dict[gate_activation]

    helper = LayerHelper('gru_unit', **locals())
    dtype = helper.input_dtype()
    size = size / 3

    # create weight
798 799
    weight = helper.create_parameter(
        attr=helper.param_attr, shape=[size, 3 * size], dtype=dtype)
Y
Yu Yang 已提交
800

801 802 803 804
    gate = helper.create_tmp_variable(dtype)
    reset_hidden_pre = helper.create_tmp_variable(dtype)
    updated_hidden = helper.create_tmp_variable(dtype)
    inputs = {'Input': input, 'HiddenPrev': hidden, 'Weight': weight}
Y
Yu Yang 已提交
805
    # create bias
806
    if helper.bias_attr:
Y
Yu Yang 已提交
807 808 809
        bias_size = [1, 3 * size]
        bias = helper.create_parameter(
            attr=helper.bias_attr, shape=bias_size, dtype=dtype, is_bias=True)
810
        inputs['Bias'] = bias
Y
Yu Yang 已提交
811 812 813

    helper.append_op(
        type='gru_unit',
814
        inputs=inputs,
Y
Yu Yang 已提交
815 816 817 818 819 820
        outputs={
            'Gate': gate,
            'ResetHiddenPrev': reset_hidden_pre,
            'Hidden': updated_hidden,
        },
        attrs={
821 822
            'activation': 2,  # tanh
            'gate_activation': 1,  # sigmoid
Y
Yu Yang 已提交
823 824 825 826 827
        })

    return updated_hidden, reset_hidden_pre, gate


Y
yuyang18 已提交
828
@templatedoc()
829
def linear_chain_crf(input, label, param_attr=None):
Y
yuyang18 已提交
830 831 832 833 834 835 836
    """
    Linear Chain CRF.

    ${comment}

    Args:
        input(${emission_type}): ${emission_comment}
D
dzhwinter 已提交
837
        input(${transition_type}): ${transition_comment}
Y
yuyang18 已提交
838 839 840 841
        label(${label_type}): ${label_comment}
        param_attr(ParamAttr): The attribute of the learnable parameter.

    Returns:
D
dzhwinter 已提交
842 843 844
        output(${emission_exps_type}): ${emission_exps_comment} \n
        output(${transition_exps_type}): ${transition_exps_comment} \n
        output(${log_likelihood_type}): ${log_likelihood_comment}
Y
yuyang18 已提交
845 846

    """
Y
Yu Yang 已提交
847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871
    helper = LayerHelper('linear_chain_crf', **locals())
    size = input.shape[1]
    transition = helper.create_parameter(
        attr=helper.param_attr,
        shape=[size + 2, size],
        dtype=helper.input_dtype())
    alpha = helper.create_tmp_variable(dtype=helper.input_dtype())
    emission_exps = helper.create_tmp_variable(dtype=helper.input_dtype())
    transition_exps = helper.create_tmp_variable(dtype=helper.input_dtype())
    log_likelihood = helper.create_tmp_variable(dtype=helper.input_dtype())
    helper.append_op(
        type='linear_chain_crf',
        inputs={"Emission": [input],
                "Transition": transition,
                "Label": label},
        outputs={
            "Alpha": [alpha],
            "EmissionExps": [emission_exps],
            "TransitionExps": transition_exps,
            "LogLikelihood": log_likelihood
        })

    return log_likelihood


Y
yuyang18 已提交
872
@templatedoc()
873
def crf_decoding(input, param_attr, label=None):
Y
yuyang18 已提交
874 875 876 877 878 879 880 881 882 883 884
    """
    ${comment}

    Args:
        input(${emission_type}): ${emission_comment}
        param_attr(ParamAttr): The parameter attribute for training.
        label(${label_type}): ${label_comment}

    Returns:
        ${viterbi_path_comment}
    """
Y
Yu Yang 已提交
885 886 887 888 889 890 891 892 893 894 895 896 897
    helper = LayerHelper('crf_decoding', **locals())
    transition = helper.get_parameter(param_attr.name)
    viterbi_path = helper.create_tmp_variable(dtype=helper.input_dtype())
    helper.append_op(
        type='crf_decoding',
        inputs={"Emission": [input],
                "Transition": transition,
                "Label": label},
        outputs={"ViterbiPath": [viterbi_path]})

    return viterbi_path


F
fengjiayi 已提交
898
def cos_sim(X, Y):
Y
Yu Yang 已提交
899 900 901
    """
    This function performs the cosine similarity between two tensors
    X and Y and returns that as the output.
902 903 904 905 906 907 908

    Args:
        X (Variable): The input X.
        Y (Variable): The input Y.
    
    Returns:
        Variable: the output of cosine(X, Y).
Y
Yu Yang 已提交
909
    """
F
fengjiayi 已提交
910
    helper = LayerHelper('cos_sim', **locals())
Y
Yu Yang 已提交
911 912 913 914 915 916 917 918 919 920 921 922 923
    out = helper.create_tmp_variable(dtype=X.dtype)
    xnorm = helper.create_tmp_variable(dtype=X.dtype)
    ynorm = helper.create_tmp_variable(dtype=X.dtype)
    helper.append_op(
        type='cos_sim',
        inputs={'X': [X],
                'Y': [Y]},
        outputs={'Out': [out],
                 'XNorm': [xnorm],
                 'YNorm': [ynorm]})
    return out


924
def dropout(x, dropout_prob, is_test=False, seed=None, name=None):
925 926 927 928 929 930 931 932 933 934
    """
    Computes dropout.

    Drop or keep each element of `x` independently. Dropout is a regularization
    technique for reducing overfitting by preventing neuron co-adaption during
    training. The dropout operator randomly set (according to the given dropout
    probability) the outputs of some units to zero, while others are remain
    unchanged.

    Args:
935 936 937 938 939 940 941 942 943
        x (Variable): The input tensor.
         dropout_prob (float): Probability of setting units to zero.
        is_test (bool): A flag indicating whether it is in test phrase or not.
        seed (int): A Python integer used to create random seeds. If this
                    parameter is set to None, a random seed is used.
                    NOTE: If an integer seed is given, always the same output
                    units will be dropped. DO NOT use a fixed seed in training.
        name (str|None): A name for this layer(optional). If set None, the layer
                         will be named automatically.
944 945 946 947 948 949 950 951 952 953 954

    Returns:
        Variable: A tensor variable.

    Examples:
        .. code-block:: python

          x = fluid.layers.data(name="data", shape=[32, 32], dtype="float32")
          droped = fluid.layers.dropout(input=x, dropout_rate=0.5)
    """

F
fengjiayi 已提交
955
    helper = LayerHelper('dropout', **locals())
956 957 958 959 960 961 962
    out = helper.create_tmp_variable(dtype=x.dtype)
    mask = helper.create_tmp_variable(dtype=x.dtype, stop_gradient=True)
    helper.append_op(
        type='dropout',
        inputs={'X': [x]},
        outputs={'Out': [out],
                 'Mask': [mask]},
963 964 965 966 967 968
        attrs={
            'dropout_prob': dropout_prob,
            'is_test': is_test,
            'fix_seed': seed is not None,
            'seed': seed if seed is not None else 0
        })
969 970 971
    return out


F
fengjiayi 已提交
972
def cross_entropy(input, label, soft_label=False):
Y
Yu Yang 已提交
973
    """
Y
Yibing Liu 已提交
974 975
    **Cross Entropy Layer**

976 977 978
    This layer computes the cross entropy between `input` and `label`. It
    supports both standard cross-entropy and soft-label cross-entropy loss
    computation.
Y
Yibing Liu 已提交
979 980

    1) One-hot cross-entropy:
F
fengjiayi 已提交
981
        `soft_label = False`, `Label[i, 0]` indicates the class index for sample i:
Y
yangyaming 已提交
982

Y
Yibing Liu 已提交
983
        .. math::
Y
yangyaming 已提交
984

Y
Yibing Liu 已提交
985 986 987
            Y[i] = -\log(X[i, Label[i]])

    2) Soft-label cross-entropy:
F
fengjiayi 已提交
988 989
        `soft_label = True`, `Label[i, j]` indicates the soft label of class j
        for sample i:
Y
Yibing Liu 已提交
990 991 992 993 994

        .. math::

            Y[i] = \sum_j{-Label[i, j] * log(X[i, j])}

Y
Yibing Liu 已提交
995
       Please make sure that in this case the summation of each row of `label`
Y
Yibing Liu 已提交
996 997 998
       equals one.

    3) One-hot cross-entropy with vecterized `label`:
F
fengjiayi 已提交
999 1000
         As a special case of 2), when each row of 'label' has only one
         non-zero element which is equal to 1, soft-label cross-entropy degenerates
Y
Yibing Liu 已提交
1001
         to a one-hot cross-entropy with one-hot label representation.
Y
yangyaming 已提交
1002

Y
Yibing Liu 已提交
1003
    Args:
Y
yangyaming 已提交
1004
        input (Variable|list):  a 2-D tensor with shape [N x D], where N is the
1005 1006 1007 1008
                                batch size and D is the number of classes. This
                                input is a probability computed by the previous
                                operator, which is almost always the result of
                                a softmax operator.
Y
yangyaming 已提交
1009
        label (Variable|list): the ground truth which is a 2-D tensor. When
1010 1011 1012 1013
                               `soft_label` is set to `False`, `label` is a
                               tensor<int64> with shape [N x 1]. When
                               `soft_label` is set to `True`, `label` is a
                               tensor<float/double> with shape [N x D].
F
fengjiayi 已提交
1014
        soft_label (bool): a flag indicating whether to
1015 1016
                                           interpretate the given labels as soft
                                           labels, default `False`.
Y
Yibing Liu 已提交
1017 1018 1019 1020 1021

    Returns:
         A 2-D tensor with shape [N x 1], the cross entropy loss.

    Raises:
1022 1023 1024 1025 1026
        `ValueError`: 1) the 1st dimension of `input` and `label` are not equal.
                      2) when `soft_label == True`, and the 2nd dimension of
                         `input` and `label` are not equal.
                      3) when `soft_label == False`, and the 2nd dimension of
                         `label` is not 1.
Y
Yibing Liu 已提交
1027 1028 1029 1030 1031 1032

    Examples:
        .. code-block:: python

          predict = fluid.layers.fc(input=net, size=classdim, act='softmax')
          cost = fluid.layers.cross_entropy(input=predict, label=label)
Y
Yu Yang 已提交
1033
    """
F
fengjiayi 已提交
1034
    helper = LayerHelper('cross_entropy', **locals())
Y
Yu Yang 已提交
1035 1036 1037 1038 1039 1040
    out = helper.create_tmp_variable(dtype=input.dtype)
    helper.append_op(
        type='cross_entropy',
        inputs={'X': [input],
                'Label': [label]},
        outputs={'Y': [out]},
F
fengjiayi 已提交
1041
        attrs={"soft_label": soft_label})
Y
Yu Yang 已提交
1042 1043 1044
    return out


F
fengjiayi 已提交
1045
def square_error_cost(input, label):
Y
Yu Yang 已提交
1046
    """
1047 1048
    **Square error cost layer**

1049 1050
    This layer accepts input predictions and target label and returns the
    squared error cost.
Y
ying 已提交
1051

1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064
    For predictions, :math:`X`, and target labels, :math:`Y`, the equation is:

    .. math::

        Out = (X - Y)^2

    In the above equation:

        * :math:`X`: Input predictions, a tensor.
        * :math:`Y`: Input labels, a tensor.
        * :math:`Out`: Output value, same shape with :math:`X`.

    Args:
1065 1066
        input (Variable): Input tensor, has predictions.
        label (Variable): Label tensor, has target labels.
1067 1068

    Returns:
G
guosheng 已提交
1069
        Variable: The tensor variable storing the element-wise squared error \
1070
                  difference of input and label.
1071 1072 1073 1074 1075 1076 1077 1078

    Examples:
        .. code-block:: python

          y = layers.data(name='y', shape=[1], dtype='float32')
          y_predict = layers.data(name='y_predict', shape=[1], dtype='float32')
          cost = layers.square_error_cost(input=y_predict, label=y)

Y
Yu Yang 已提交
1079
    """
F
fengjiayi 已提交
1080
    helper = LayerHelper('square_error_cost', **locals())
Y
Yu Yang 已提交
1081 1082 1083 1084 1085 1086 1087 1088 1089
    minus_out = helper.create_tmp_variable(dtype=input.dtype)
    helper.append_op(
        type='elementwise_sub',
        inputs={'X': [input],
                'Y': [label]},
        outputs={'Out': [minus_out]})

    square_out = helper.create_tmp_variable(dtype=input.dtype)
    helper.append_op(
F
fengjiayi 已提交
1090 1091
        type='square', inputs={'X': [minus_out]},
        outputs={'Out': [square_out]})
Y
Yu Yang 已提交
1092 1093 1094
    return square_out


1095
@templatedoc()
Y
Yu Yang 已提交
1096 1097 1098 1099
def chunk_eval(input,
               label,
               chunk_scheme,
               num_chunk_types,
F
fengjiayi 已提交
1100
               excluded_chunk_types=None):
Y
Yu Yang 已提交
1101
    """
Y
yangyaming 已提交
1102
    This function computes and outputs the precision, recall and
1103
    F1-score of chunk detection.
1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115

    Args:
        input (Variable): prediction output of the network.
        label (Variable): label of the test data set.
        chunk_scheme (str): ${chunk_scheme_comment}
        num_chunk_types (int): ${num_chunk_types_comment}
        excluded_chunk_types (list): ${excluded_chunk_types_comment}
    
    Returns:
        tuple: tuple containing: (precision, recall, f1_score,
               num_infer_chunks, num_label_chunks,
               num_correct_chunks)
Y
Yu Yang 已提交
1116
    """
F
fengjiayi 已提交
1117
    helper = LayerHelper("chunk_eval", **locals())
Y
Yu Yang 已提交
1118 1119 1120 1121 1122

    # prepare output
    precision = helper.create_tmp_variable(dtype="float32")
    recall = helper.create_tmp_variable(dtype="float32")
    f1_score = helper.create_tmp_variable(dtype="float32")
1123 1124 1125
    num_infer_chunks = helper.create_tmp_variable(dtype="int64")
    num_label_chunks = helper.create_tmp_variable(dtype="int64")
    num_correct_chunks = helper.create_tmp_variable(dtype="int64")
Y
Yu Yang 已提交
1126 1127 1128 1129 1130 1131 1132 1133

    helper.append_op(
        type="chunk_eval",
        inputs={"Inference": [input],
                "Label": [label]},
        outputs={
            "Precision": [precision],
            "Recall": [recall],
1134 1135 1136 1137
            "F1-Score": [f1_score],
            "NumInferChunks": [num_infer_chunks],
            "NumLabelChunks": [num_label_chunks],
            "NumCorrectChunks": [num_correct_chunks]
Y
Yu Yang 已提交
1138 1139 1140
        },
        attrs={
            "num_chunk_types": num_chunk_types,
G
guosheng 已提交
1141 1142
            "chunk_scheme": chunk_scheme,
            "excluded_chunk_types": excluded_chunk_types or []
Y
Yu Yang 已提交
1143
        })
1144 1145
    return (precision, recall, f1_score, num_infer_chunks, num_label_chunks,
            num_correct_chunks)
Y
Yu Yang 已提交
1146 1147


1148
@templatedoc()
Y
Yu Yang 已提交
1149 1150 1151 1152 1153 1154 1155
def sequence_conv(input,
                  num_filters,
                  filter_size=3,
                  filter_stride=1,
                  padding=None,
                  bias_attr=None,
                  param_attr=None,
1156
                  act=None):
Y
Yu Yang 已提交
1157 1158 1159 1160
    """
    This function creates the op for sequence_conv, using the inputs and
    other convolutional configurations for the filters and stride as given
    in the input parameters to the function.
1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173

    Args:
        input (Variable): ${x_comment}
        num_filters (int): number of filters.
        filter_size (int): the filter size (H and W).
        filter_stride (int): stride of the filter.
        padding (bool): if True, add paddings.
        bias_attr (ParamAttr|None): attributes for bias
        param_attr (ParamAttr|None): attributes for parameter
        act (str): the activation type
    
    Returns:
        Variable: output of sequence_conv
Y
Yu Yang 已提交
1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198
    """

    helper = LayerHelper('sequence_conv', **locals())
    dtype = helper.input_dtype()
    filter_shape = [filter_size * input.shape[1], num_filters]
    filter_param = helper.create_parameter(
        attr=helper.param_attr, shape=filter_shape, dtype=dtype)
    pre_bias = helper.create_tmp_variable(dtype)

    helper.append_op(
        type='sequence_conv',
        inputs={
            'X': [input],
            'Filter': [filter_param],
        },
        outputs={"Out": pre_bias},
        attrs={
            'contextStride': filter_stride,
            'contextStart': -int(filter_size / 2),
            'contextLength': filter_size
        })
    pre_act = helper.append_bias_op(pre_bias)
    return helper.append_activation(pre_act)


1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210
def sequence_softmax(input, param_attr=None, bias_attr=None, use_cudnn=True):
    helper = LayerHelper('sequence_softmax', **locals())
    dtype = helper.input_dtype()
    softmax_out = helper.create_tmp_variable(dtype)
    helper.append_op(
        type="sequence_softmax",
        inputs={"X": input},
        outputs={"Out": softmax_out},
        attrs={"use_cudnn": use_cudnn})
    return softmax_out


1211
def softmax(input, param_attr=None, bias_attr=None, use_cudnn=True, name=None):
1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222
    helper = LayerHelper('softmax', **locals())
    dtype = helper.input_dtype()
    softmax_out = helper.create_tmp_variable(dtype)
    helper.append_op(
        type="softmax",
        inputs={"X": input},
        outputs={"Out": softmax_out},
        attrs={"use_cudnn": use_cudnn})
    return softmax_out


Y
Yu Yang 已提交
1223 1224 1225
def conv2d(input,
           num_filters,
           filter_size,
C
chengduoZH 已提交
1226 1227
           stride=1,
           padding=0,
1228
           dilation=1,
Y
Yu Yang 已提交
1229 1230 1231
           groups=None,
           param_attr=None,
           bias_attr=None,
C
chengduoZH 已提交
1232
           use_cudnn=True,
1233
           use_mkldnn=False,
1234 1235
           act=None,
           name=None):
Y
Yu Yang 已提交
1236
    """
C
chengduoZH 已提交
1237 1238 1239
    **Convlution2D Layer**

    The convolution2D layer calculates the output based on the input, filter
1240 1241 1242
    and strides, paddings, dilations, groups parameters. Input(Input) and
    Output(Output) are in NCHW format. Where N is batch size, C is the number of
    channels, H is the height of the feature, and W is the width of the feature.
C
chengduoZH 已提交
1243 1244
    The details of convolution layer, please refer UFLDL's `convolution,
    <http://ufldl.stanford.edu/tutorial/supervised/FeatureExtractionUsingConvolution/>`_ .
1245 1246 1247
    If bias attribution and activation type are provided, bias is added to the
    output of the convolution, and the corresponding activation function is
    applied to the final result.
C
chengduoZH 已提交
1248

1249
    For each input :math:`X`, the equation is:
C
refine  
chengduoZH 已提交
1250

C
chengduoZH 已提交
1251 1252
    .. math::

C
refine  
chengduoZH 已提交
1253
        Out = \sigma (W \\ast X + b)
C
chengduoZH 已提交
1254

C
chengduoZH 已提交
1255
    In the above equation:
C
chengduoZH 已提交
1256

1257 1258 1259 1260 1261
    * :math:`X`: Input value, a tensor with NCHW format.
    * :math:`W`: Filter value, a tensor with MCHW format.
    * :math:`\\ast`: Convolution operation.
    * :math:`b`: Bias value, a 2-D tensor with shape [M, 1].
    * :math:`\\sigma`: Activation function.
1262 1263
    * :math:`Out`: Output value, the shape of :math:`Out` and :math:`X` may be
                   different.
C
chengduoZH 已提交
1264 1265 1266

    Example:

1267 1268
        - Input:

W
weixing02 已提交
1269
          Input shape: :math:`(N, C_{in}, H_{in}, W_{in})`
C
refine  
chengduoZH 已提交
1270

W
weixing02 已提交
1271
          Filter shape: :math:`(C_{out}, C_{in}, H_f, W_f)`
C
refine  
chengduoZH 已提交
1272

1273
        - Output:
W
weixing02 已提交
1274
          Output shape: :math:`(N, C_{out}, H_{out}, W_{out})`
C
refine  
chengduoZH 已提交
1275

C
chengduoZH 已提交
1276
        Where
1277 1278

        .. math::
C
chengduoZH 已提交
1279

W
weixing02 已提交
1280 1281
            H_{out}&= \\frac{(H_{in} + 2 * paddings[0] - (dilations[0] * (H_f - 1) + 1))}{strides[0]} + 1 \\\\
            W_{out}&= \\frac{(W_{in} + 2 * paddings[1] - (dilations[1] * (W_f - 1) + 1))}{strides[1]} + 1
C
chengduoZH 已提交
1282 1283

    Args:
1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311
        input (Variable): The input image with [N, C, H, W] format.
            num_filters(int): The number of filter. It is as same as the output
            image channel.
        filter_size (int|tuple|None): The filter size. If filter_size is a tuple,
            it must contain two integers, (filter_size_H, filter_size_W).
            Otherwise, the filter will be a square.
        stride (int|tuple): The stride size. If stride is a tuple, it must
            contain two integers, (stride_H, stride_W). Otherwise, the
            stride_H = stride_W = stride. Default: stride = 1.
        padding (int|tuple): The padding size. If padding is a tuple, it must
            contain two integers, (padding_H, padding_W). Otherwise, the
            padding_H = padding_W = padding. Default: padding = 0.
        dilation (int|tuple): The dilation size. If dilation is a tuple, it must
            contain two integers, (dilation_H, dilation_W). Otherwise, the
            dilation_H = dilation_W = dilation. Default: dilation = 1.
        groups (int): The groups number of the Conv2d Layer. According to grouped
            convolution in Alex Krizhevsky's Deep CNN paper: when group=2,
            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
            connected to the second half of the input channels. Default: groups=1
        param_attr (ParamAttr): The parameters to the Conv2d Layer. Default: None
        bias_attr (ParamAttr): Bias parameter for the Conv2d layer. Default: None
        use_cudnn (bool): Use cudnn kernel or not, it is valid only when the cudnn
            library is installed. Default: True
        use_mkldnn (bool): Use mkldnn kernels or not.
        act (str): Activation type. Default: None
        name (str|None): A name for this layer(optional). If set None, the layer
            will be named automatically.
C
chengduoZH 已提交
1312 1313

    Returns:
G
guosheng 已提交
1314
        Variable: The tensor variable storing the convolution and \
C
chengduoZH 已提交
1315 1316
                  non-linearity activation result.

C
refine  
chengduoZH 已提交
1317
    Raises:
1318 1319
        ValueError: If the shapes of input, filter_size, stride, padding and
                    groups mismatch.
C
refine  
chengduoZH 已提交
1320

C
chengduoZH 已提交
1321 1322 1323
    Examples:
        .. code-block:: python

1324 1325 1326 1327
          data = fluid.layers.data(
              name='data', shape=[3, 32, 32], dtype='float32')
          conv2d = fluid.layers.conv2d(
              input=data, num_filters=2, filter_size=3, act="relu")
Y
Yu Yang 已提交
1328 1329 1330
    """

    num_channels = input.shape[1]
1331 1332

    l_type = 'conv2d'
X
xzl 已提交
1333 1334
    if (num_channels == groups and num_filters % num_channels == 0 and
            not use_cudnn):
1335
        l_type = 'depthwise_conv2d'
1336 1337 1338 1339

    helper = LayerHelper(l_type, **locals())
    dtype = helper.input_dtype()

Y
Yu Yang 已提交
1340 1341 1342 1343 1344 1345 1346
    if groups is None:
        num_filter_channels = num_channels
    else:
        if num_channels % groups != 0:
            raise ValueError("num_channels must be divisible by groups.")
        num_filter_channels = num_channels / groups

C
chengduoZH 已提交
1347 1348 1349
    filter_size = utils.convert_to_list(filter_size, 2, 'filter_size')
    stride = utils.convert_to_list(stride, 2, 'stride')
    padding = utils.convert_to_list(padding, 2, 'padding')
1350
    dilation = utils.convert_to_list(dilation, 2, 'dilation')
C
chengduoZH 已提交
1351

C
chengduoZH 已提交
1352 1353
    if not isinstance(use_cudnn, bool):
        raise ValueError("use_cudnn should be True or False")
Y
Yu Yang 已提交
1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370

    input_shape = input.shape
    filter_shape = [num_filters, num_filter_channels] + filter_size

    def _get_default_param_initializer():
        std = (2.0 / (filter_size[0]**2 * num_channels))**0.5
        return Normal(0.0, std, 0)

    filter_param = helper.create_parameter(
        attr=helper.param_attr,
        shape=filter_shape,
        dtype=dtype,
        default_initializer=_get_default_param_initializer())

    pre_bias = helper.create_tmp_variable(dtype)

    helper.append_op(
1371
        type=l_type,
Y
Yu Yang 已提交
1372 1373 1374 1375 1376
        inputs={
            'Input': input,
            'Filter': filter_param,
        },
        outputs={"Output": pre_bias},
C
chengduoZH 已提交
1377 1378 1379
        attrs={
            'strides': stride,
            'paddings': padding,
1380
            'dilations': dilation,
C
chengduoZH 已提交
1381
            'groups': groups,
1382 1383
            'use_cudnn': use_cudnn,
            'use_mkldnn': use_mkldnn
C
chengduoZH 已提交
1384
        })
Y
Yu Yang 已提交
1385 1386 1387 1388 1389 1390

    pre_act = helper.append_bias_op(pre_bias, dim_start=1, dim_end=2)

    return helper.append_activation(pre_act)


C
chengduoZH 已提交
1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408
def conv3d(input,
           num_filters,
           filter_size,
           stride=1,
           padding=0,
           dilation=1,
           groups=None,
           param_attr=None,
           bias_attr=None,
           use_cudnn=True,
           use_mkldnn=False,
           act=None,
           name=None):
    """
    **Convlution3D Layer**

    The convolution3D layer calculates the output based on the input, filter
    and strides, paddings, dilations, groups parameters. Input(Input) and
1409 1410 1411 1412 1413 1414
    Output(Output) are in NCDHW 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 is the width of the feature. Convlution3D is similar with Convlution2D
    but adds one dimension(depth). If bias attribution and activation type are
    provided, bias is added to the output of the convolution, and the
    corresponding activation function is applied to the final result.
C
chengduoZH 已提交
1415 1416 1417 1418 1419 1420 1421 1422 1423

    For each input :math:`X`, the equation is:

    .. math::

        Out = \sigma (W \\ast X + b)

    In the above equation:

1424 1425
    * :math:`X`: Input value, a tensor with NCDHW format.
    * :math:`W`: Filter value, a tensor with MCDHW format.
C
chengduoZH 已提交
1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455
    * :math:`\\ast`: Convolution operation.
    * :math:`b`: Bias value, a 2-D tensor with shape [M, 1].
    * :math:`\\sigma`: Activation function.
    * :math:`Out`: Output value, the shape of :math:`Out` and :math:`X` may be
                   different.

    Example:

        - Input:

          Input shape: :math:`(N, C_{in}, D_{in}, H_{in}, W_{in})`

          Filter shape: :math:`(C_{out}, C_{in}, D_f, H_f, W_f)`

        - Output:
          Output shape: :math:`(N, C_{out}, D_{out}, H_{out}, W_{out})`

        Where

        .. math::

            D_{out}&= \\frac{(D_{in} + 2 * paddings[0] - (dilations[0] * (D_f - 1) + 1))}{strides[0]} + 1 \\\\
            H_{out}&= \\frac{(H_{in} + 2 * paddings[1] - (dilations[1] * (H_f - 1) + 1))}{strides[1]} + 1 \\\\
            W_{out}&= \\frac{(W_{in} + 2 * paddings[2] - (dilations[2] * (W_f - 1) + 1))}{strides[2]} + 1

    Args:
        input (Variable): The input image with [N, C, D, H, W] format.
            num_filters(int): The number of filter. It is as same as the output
            image channel.
        filter_size (int|tuple|None): The filter size. If filter_size is a tuple,
1456
            it must contain three integers, (filter_size_D, filter_size_H, filter_size_W).
C
chengduoZH 已提交
1457 1458
            Otherwise, the filter will be a square.
        stride (int|tuple): The stride size. If stride is a tuple, it must
1459
            contain three integers, (stride_D, stride_H, stride_W). Otherwise, the
C
chengduoZH 已提交
1460 1461
            stride_D = stride_H = stride_W = stride. Default: stride = 1.
        padding (int|tuple): The padding size. If padding is a tuple, it must
1462
            contain three integers, (padding_D, padding_H, padding_W). Otherwise, the
C
chengduoZH 已提交
1463 1464
            padding_D = padding_H = padding_W = padding. Default: padding = 0.
        dilation (int|tuple): The dilation size. If dilation is a tuple, it must
1465
            contain three integers, (dilation_D, dilation_H, dilation_W). Otherwise, the
C
chengduoZH 已提交
1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550
            dilation_D = dilation_H = dilation_W = dilation. Default: dilation = 1.
        groups (int): The groups number of the Conv3d Layer. According to grouped
            convolution in Alex Krizhevsky's Deep CNN paper: when group=2,
            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
            connected to the second half of the input channels. Default: groups=1
        param_attr (ParamAttr): The parameters to the Conv3d Layer. Default: None
        bias_attr (ParamAttr): Bias parameter for the Conv3d layer. Default: None
        use_cudnn (bool): Use cudnn kernel or not, it is valid only when the cudnn
            library is installed. Default: True
        use_mkldnn (bool): Use mkldnn kernels or not.
        act (str): Activation type. Default: None
        name (str|None): A name for this layer(optional). If set None, the layer
            will be named automatically.

    Returns:
        Variable: The tensor variable storing the convolution and \
                  non-linearity activation result.

    Raises:
        ValueError: If the shapes of input, filter_size, stride, padding and
                    groups mismatch.

    Examples:
        .. code-block:: python

          data = fluid.layers.data(
              name='data', shape=[3, 12, 32, 32], dtype='float32')
          conv2d = fluid.layers.conv3d(
              input=data, num_filters=2, filter_size=3, act="relu")
    """

    l_type = 'conv3d'

    helper = LayerHelper(l_type, **locals())
    dtype = helper.input_dtype()

    num_channels = input.shape[1]

    if groups is None:
        num_filter_channels = num_channels
    else:
        if num_channels % groups != 0:
            raise ValueError("num_channels must be divisible by groups.")
        num_filter_channels = num_channels / groups

    filter_size = utils.convert_to_list(filter_size, 3, 'filter_size')
    stride = utils.convert_to_list(stride, 3, 'stride')
    padding = utils.convert_to_list(padding, 3, 'padding')
    dilation = utils.convert_to_list(dilation, 3, 'dilation')

    if not isinstance(use_cudnn, bool):
        raise ValueError("use_cudnn should be True or False")

    input_shape = input.shape
    filter_shape = [num_filters, num_filter_channels] + filter_size

    def _get_default_param_initializer():
        std = (2.0 / (filter_size[0]**3 * num_channels))**0.5
        return Normal(0.0, std, 0)

    filter_param = helper.create_parameter(
        attr=helper.param_attr,
        shape=filter_shape,
        dtype=dtype,
        default_initializer=_get_default_param_initializer())

    pre_bias = helper.create_tmp_variable(dtype)

    helper.append_op(
        type=l_type,
        inputs={
            'Input': input,
            'Filter': filter_param,
        },
        outputs={"Output": pre_bias},
        attrs={
            'strides': stride,
            'paddings': padding,
            'dilations': dilation,
            'groups': groups,
            'use_cudnn': use_cudnn,
            'use_mkldnn': use_mkldnn
        })

1551
    pre_act = helper.append_bias_op(pre_bias, dim_start=1, dim_end=2)
C
chengduoZH 已提交
1552 1553 1554 1555

    return helper.append_activation(pre_act)


F
fengjiayi 已提交
1556
def sequence_pool(input, pool_type):
Y
Yu Yang 已提交
1557
    """
Y
yangyaming 已提交
1558 1559 1560
    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.
L
Luo Tao 已提交
1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585

    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)
1586 1587
         last   : out.data = [3, 6, 1], where 3=last(1,3), 6=last(2,4,6), 1=last(5,1)
         first  : out.data = [1, 2, 5], where 1=first(1,3), 2=first(2,4,6), 5=first(5,1)
F
fengjiayi 已提交
1588

L
Luo Tao 已提交
1589 1590
    Args:
        input(variable): The input variable which is a LoDTensor.
Y
yangyaming 已提交
1591
        pool_type (string): The pooling type of sequence_pool.
L
Luo Tao 已提交
1592 1593 1594 1595 1596 1597 1598 1599
            It supports average, sum, sqrt and max.

    Returns:
        The sequence pooling variable which is a Tensor.

    Examples:

        .. code-block:: python
F
fengjiayi 已提交
1600

Y
yangyaming 已提交
1601
             x = fluid.layers.data(name='x', shape=[7, 1],
L
Luo Tao 已提交
1602 1603 1604 1605 1606
                              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')
1607 1608
             last_x = fluid.layers.sequence_pool(input=x, pool_type='last')
             first_x = fluid.layers.sequence_pool(input=x, pool_type='first')
Y
Yu Yang 已提交
1609
    """
F
fengjiayi 已提交
1610
    helper = LayerHelper('sequence_pool', **locals())
Y
Yu Yang 已提交
1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621
    dtype = helper.input_dtype()
    pool_out = helper.create_tmp_variable(dtype)
    max_index = helper.create_tmp_variable(dtype)

    helper.append_op(
        type="sequence_pool",
        inputs={"X": input},
        outputs={"Out": pool_out,
                 "MaxIndex": max_index},
        attrs={"pooltype": pool_type.upper()})

Y
yangyaming 已提交
1622 1623 1624 1625 1626
    # when pool_type is max, variable max_index is initialized,
    # so we stop the gradient explicitly here
    if pool_type == 'max':
        max_index.stop_gradient = True

Y
Yu Yang 已提交
1627 1628 1629
    return pool_out


F
fengjiayi 已提交
1630
def sequence_first_step(input):
L
Luo Tao 已提交
1631
    """
L
Luo Tao 已提交
1632
    This function gets the first step of sequence.
L
Luo Tao 已提交
1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644

    .. 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)
F
fengjiayi 已提交
1645

L
Luo Tao 已提交
1646 1647 1648 1649 1650 1651 1652 1653 1654
    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
F
fengjiayi 已提交
1655

Y
yangyaming 已提交
1656
             x = fluid.layers.data(name='x', shape=[7, 1],
L
Luo Tao 已提交
1657 1658 1659
                              dtype='float32', lod_level=1)
             x_first_step = fluid.layers.sequence_first_step(input=x)
    """
1660 1661 1662
    return sequence_pool(input=input, pool_type="first")


F
fengjiayi 已提交
1663
def sequence_last_step(input):
L
Luo Tao 已提交
1664
    """
L
Luo Tao 已提交
1665
    This function gets the last step of sequence.
L
Luo Tao 已提交
1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677

    .. 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)
F
fengjiayi 已提交
1678

L
Luo Tao 已提交
1679 1680 1681 1682 1683 1684 1685 1686 1687
    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
F
fengjiayi 已提交
1688

Y
yangyaming 已提交
1689
             x = fluid.layers.data(name='x', shape=[7, 1],
L
Luo Tao 已提交
1690 1691 1692
                              dtype='float32', lod_level=1)
             x_last_step = fluid.layers.sequence_last_step(input=x)
    """
1693 1694 1695
    return sequence_pool(input=input, pool_type="last")


Y
Yu Yang 已提交
1696
def pool2d(input,
C
chengduoZH 已提交
1697 1698
           pool_size=-1,
           pool_type="max",
C
chengduoZH 已提交
1699 1700
           pool_stride=1,
           pool_padding=0,
C
caoying03 已提交
1701
           global_pooling=False,
C
chengduoZH 已提交
1702
           use_cudnn=True,
1703
           ceil_mode=False,
1704
           use_mkldnn=False,
C
caoying03 已提交
1705
           name=None):
Y
Yu Yang 已提交
1706 1707 1708
    """
    This function adds the operator for pooling in 2 dimensions, using the
    pooling configurations mentioned in input parameters.
1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724

    Args:
        input (Variable): ${input_comment}
        pool_size (int): ${ksize_comment}
        pool_type (str): ${pooling_type_comment}
        pool_stride (int): stride of the pooling layer.
        pool_padding (int): padding size.
        global_pooling (bool): ${global_pooling_comment}
        use_cudnn (bool): ${use_cudnn_comment}
        ceil_mode (bool): ${ceil_mode_comment}
        use_mkldnn (bool): ${use_mkldnn_comment}
        name (str): A name for this layer(optional). If set None, the layer
            will be named automatically.
    
    Returns:
        Variable: output of pool2d layer.
Y
Yu Yang 已提交
1725 1726 1727 1728 1729
    """
    if pool_type not in ["max", "avg"]:
        raise ValueError(
            "Unknown pool_type: '%s'. It can only be 'max' or 'avg'.",
            str(pool_type))
C
chengduoZH 已提交
1730

C
chengduoZH 已提交
1731 1732 1733 1734 1735
    if global_pooling is False and pool_size == -1:
        raise ValueError(
            "When the global_pooling is False, pool_size must be passed "
            "and be a valid value. Received pool_size: " + str(pool_size))

C
chengduoZH 已提交
1736 1737 1738 1739
    pool_size = utils.convert_to_list(pool_size, 2, 'pool_size')
    pool_padding = utils.convert_to_list(pool_padding, 2, 'pool_padding')
    pool_stride = utils.convert_to_list(pool_stride, 2, 'pool_stride')

C
chengduoZH 已提交
1740 1741
    if not isinstance(use_cudnn, bool):
        raise ValueError("use_cudnn should be True or False")
Y
Yu Yang 已提交
1742

C
Add doc  
chengduoZH 已提交
1743
    l_type = 'pool2d'
1744 1745

    helper = LayerHelper(l_type, **locals())
Y
Yu Yang 已提交
1746 1747 1748 1749
    dtype = helper.input_dtype()
    pool_out = helper.create_tmp_variable(dtype)

    helper.append_op(
1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820
        type=l_type,
        inputs={"X": input},
        outputs={"Out": pool_out},
        attrs={
            "pooling_type": pool_type,
            "ksize": pool_size,
            "global_pooling": global_pooling,
            "strides": pool_stride,
            "paddings": pool_padding,
            "use_cudnn": use_cudnn,
            "ceil_mode": ceil_mode,
            "use_mkldnn": use_mkldnn
        })

    return pool_out


def pool3d(input,
           pool_size=-1,
           pool_type="max",
           pool_stride=1,
           pool_padding=0,
           global_pooling=False,
           use_cudnn=True,
           ceil_mode=False,
           use_mkldnn=False,
           name=None):
    """
    This function adds the operator for pooling in 3-dimensions, using the
    pooling configurations mentioned in input parameters.

    Args:
        input (Variable): ${input_comment}
        pool_size (int): ${ksize_comment}
        pool_type (str): ${pooling_type_comment}
        pool_stride (int): stride of the pooling layer.
        pool_padding (int): padding size.
        global_pooling (bool): ${global_pooling_comment}
        use_cudnn (bool): ${use_cudnn_comment}
        ceil_mode (bool): ${ceil_mode_comment}
        use_mkldnn (bool): ${use_mkldnn_comment}
        name (str): A name for this layer(optional). If set None, the layer
            will be named automatically.

    Returns:
        Variable: output of pool3d layer.
    """
    if pool_type not in ["max", "avg"]:
        raise ValueError(
            "Unknown pool_type: '%s'. It can only be 'max' or 'avg'.",
            str(pool_type))

    if global_pooling is False and pool_size == -1:
        raise ValueError(
            "When the global_pooling is False, pool_size must be passed "
            "and be a valid value. Received pool_size: " + str(pool_size))

    pool_size = utils.convert_to_list(pool_size, 3, 'pool_size')
    pool_padding = utils.convert_to_list(pool_padding, 3, 'pool_padding')
    pool_stride = utils.convert_to_list(pool_stride, 3, 'pool_stride')

    if not isinstance(use_cudnn, bool):
        raise ValueError("use_cudnn should be True or False")

    l_type = "pool3d"
    helper = LayerHelper(l_type, **locals())
    dtype = helper.input_dtype()
    pool_out = helper.create_tmp_variable(dtype)

    helper.append_op(
        type=l_type,
Y
Yu Yang 已提交
1821 1822 1823 1824 1825 1826 1827
        inputs={"X": input},
        outputs={"Out": pool_out},
        attrs={
            "pooling_type": pool_type,
            "ksize": pool_size,
            "global_pooling": global_pooling,
            "strides": pool_stride,
C
chengduoZH 已提交
1828
            "paddings": pool_padding,
1829
            "use_cudnn": use_cudnn,
1830 1831
            "ceil_mode": ceil_mode,
            "use_mkldnn": use_mkldnn
Y
Yu Yang 已提交
1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843
        })

    return pool_out


def batch_norm(input,
               act=None,
               is_test=False,
               momentum=0.9,
               epsilon=1e-05,
               param_attr=None,
               bias_attr=None,
C
caoying03 已提交
1844
               data_layout='NCHW',
Y
Yang Yang 已提交
1845
               in_place=False,
1846
               use_mkldnn=False,
1847 1848
               name=None,
               moving_mean_name=None,
W
wanghaoshuang 已提交
1849
               moving_variance_name=None,
W
wanghaoshuang 已提交
1850
               do_model_average_for_mean_and_var=False):
Y
Yu Yang 已提交
1851 1852 1853
    """
    This function helps create an operator to implement
    the BatchNorm layer using the configurations from the input parameters.
1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872

    Args:
        input (Variable): the input variable.
        act (str): activation type
        is_test (bool): whether to run batch_norm as test mode.
        momentum (float): momentum
        epsilon (float): epsilon, default 1e-05
        param_attr (ParamAttr|None): attributes for parameter
        bias_attr (ParamAttr|None): attributes for bias
        data_layout (str): data layout, default NCHW
        in_place (bool): if True, do not create tmp variable
        use_mkldnn (bool): ${use_mkldnn_comment}
        name (str): The name of this layer. It is optional.
        moving_mean_name (str): The name of moving mean variable name, optional.
        moving_variance_name (str): The name of moving variance name, optional.
        do_model_average_for_mean_and_var (bool):

    Returns:
        Variable: output of batch_norm layer.
Y
Yu Yang 已提交
1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895
    """
    helper = LayerHelper('batch_norm', **locals())
    dtype = helper.input_dtype()

    input_shape = input.shape
    if data_layout == 'NCHW':
        channel_num = input_shape[1]
    else:
        if data_layout == 'NHWC':
            channel_num = input_shape[-1]
        else:
            raise ValueError("unsupported data layout:" + data_layout)

    param_shape = [channel_num]

    # create parameter
    scale = helper.create_parameter(
        attr=helper.param_attr,
        shape=param_shape,
        dtype=dtype,
        default_initializer=Constant(1.0))

    bias = helper.create_parameter(
1896
        attr=helper.bias_attr, shape=param_shape, dtype=dtype, is_bias=True)
Y
Yu Yang 已提交
1897

1898 1899
    mean = helper.create_parameter(
        attr=ParamAttr(
W
wanghaoshuang 已提交
1900 1901 1902
            name=moving_mean_name,
            initializer=Constant(0.0),
            trainable=False,
W
wanghaoshuang 已提交
1903
            do_model_average=do_model_average_for_mean_and_var),
Q
QI JUN 已提交
1904
        shape=param_shape,
1905 1906 1907 1908 1909 1910 1911
        dtype=input.dtype)
    mean.stop_gradient = True

    variance = helper.create_parameter(
        attr=ParamAttr(
            name=moving_variance_name,
            initializer=Constant(1.0),
W
wanghaoshuang 已提交
1912
            trainable=False,
W
wanghaoshuang 已提交
1913
            do_model_average=do_model_average_for_mean_and_var),
Q
QI JUN 已提交
1914
        shape=param_shape,
1915 1916
        dtype=input.dtype)
    variance.stop_gradient = True
Y
Yu Yang 已提交
1917 1918 1919 1920 1921 1922

    # create output
    # mean and mean_out share the same memory
    mean_out = mean
    # variance and variance out share the same memory
    variance_out = variance
Q
QI JUN 已提交
1923 1924
    saved_mean = helper.create_tmp_variable(dtype=dtype, stop_gradient=True)
    saved_variance = helper.create_tmp_variable(dtype=dtype, stop_gradient=True)
Y
Yu Yang 已提交
1925

Y
Yang Yang 已提交
1926
    batch_norm_out = input if in_place else helper.create_tmp_variable(dtype)
Y
Yu Yang 已提交
1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943

    helper.append_op(
        type="batch_norm",
        inputs={
            "X": input,
            "Scale": scale,
            "Bias": bias,
            "Mean": mean,
            "Variance": variance
        },
        outputs={
            "Y": batch_norm_out,
            "MeanOut": mean_out,
            "VarianceOut": variance_out,
            "SavedMean": saved_mean,
            "SavedVariance": saved_variance
        },
1944 1945 1946 1947 1948 1949
        attrs={
            "momentum": momentum,
            "epsilon": epsilon,
            "is_test": is_test,
            "use_mkldnn": use_mkldnn
        })
Y
Yu Yang 已提交
1950 1951 1952 1953

    return helper.append_activation(batch_norm_out)


G
guosheng 已提交
1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965
def layer_norm(input,
               scale=True,
               shift=True,
               begin_norm_axis=1,
               epsilon=1e-05,
               param_attr=None,
               bias_attr=None,
               act=None,
               name=None):
    """
    **Layer Normalization**

1966
    Assume feature vectors exist on dimensions
G
guosheng 已提交
1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986
    :attr:`begin_norm_axis ... rank(input)` and calculate the moment statistics
    along these dimensions for each feature vector :math:`a` with size
    :math:`H`, then normalize each feature vector using the corresponding
    statistics. After that, apply learnable gain and bias on the normalized
    tensor to scale and shift if :attr:`scale` and :attr:`shift` are set.

    Refer to `Layer Normalization <https://arxiv.org/pdf/1607.06450v1.pdf>`_

    The formula is as follows:

    .. math::

        \\mu & = \\frac{1}{H}\\sum_{i=1}^{H} a_i

        \\sigma & = \\sqrt{\\frac{1}{H}\sum_{i=1}^{H}(a_i - \\mu)^2}

        h & = f(\\frac{g}{\\sigma}(a - \\mu) + b)

    Args:
        input(Variable): The input tensor variable.
1987
        scale(bool): Whether to learn the adaptive gain :math:`g` after
G
guosheng 已提交
1988
            normalization.
1989
        shift(bool): Whether to learn the adaptive bias :math:`b` after
G
guosheng 已提交
1990
            normalization.
1991
        begin_norm_axis(bool): The normalization will be performed along
G
guosheng 已提交
1992
            dimensions from :attr:`begin_norm_axis` to :attr:`rank(input)`.
1993
        epsilon(float): The small value added to the variance to prevent
G
guosheng 已提交
1994 1995 1996 1997 1998 1999
            division by zero.
        param_attr(ParamAttr|None): The parameter attribute for the learnable
            gain :math:`g`.
        bias_attr(ParamAttr|None): The parameter attribute for the learnable
            bias :math:`b`.
        act(str): Activation to be applied to the output of layer normalizaiton.
2000
        name (str): The name of this layer. It is optional.
G
guosheng 已提交
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025

    Returns:
        Variable: A tensor variable with the same shape as the input.

    Examples:
        .. code-block:: python

            data = fluid.layers.data(
              name='data', shape=[3, 32, 32], dtype='float32')
            x = fluid.layers.layer_norm(input=data, begin_norm_axis=1)
    """
    helper = LayerHelper('layer_norm', **locals())
    dtype = helper.input_dtype()

    # create intput and parameters
    inputs = {'X': input}
    input_shape = input.shape
    param_shape = [reduce(lambda x, y: x * y, input_shape[begin_norm_axis:])]
    if scale:
        scale = helper.create_parameter(
            attr=helper.param_attr,
            shape=param_shape,
            dtype=dtype,
            default_initializer=Constant(1.0))
        inputs['Scale'] = scale
G
guosheng 已提交
2026
    if shift:
G
guosheng 已提交
2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050
        assert bias_attr is not False
        bias = helper.create_parameter(
            attr=helper.bias_attr, shape=param_shape, dtype=dtype, is_bias=True)
        inputs['Bias'] = bias

    # create output
    mean_out = helper.create_tmp_variable(dtype=dtype, stop_gradient=True)
    variance_out = helper.create_tmp_variable(dtype=dtype, stop_gradient=True)
    layer_norm_out = helper.create_tmp_variable(dtype)

    helper.append_op(
        type="layer_norm",
        inputs=inputs,
        outputs={
            "Y": layer_norm_out,
            "Mean": mean_out,
            "Variance": variance_out,
        },
        attrs={"epsilon": epsilon,
               "begin_norm_axis": begin_norm_axis})

    return helper.append_activation(layer_norm_out)


C
caoying03 已提交
2051
def beam_search_decode(ids, scores, name=None):
2052
    """
D
dzhwinter 已提交
2053 2054
    Beam Search Decode

D
dzhwinter 已提交
2055 2056
    This layers is to pack the output of beam search layer into sentences and
    associated scores. It is usually called after the beam search layer.
D
dzhwinter 已提交
2057 2058 2059 2060 2061 2062
    Typically, the output of beam search layer is a tensor of selected ids, with
    a tensor of the score of each id. Beam search layer's output ids, however, 
    are generated directly during the tree search, and they are stacked by each 
    level of the search tree. Thus we need to reorganize them into sentences, 
    based on the score of each id. This layer takes the output of beam search
    layer as input and repack them into sentences.
D
dzhwinter 已提交
2063

2064 2065 2066
    ${beam_search_decode}

    Args:
D
dzhwinter 已提交
2067 2068 2069
        ids (Variable): The selected ids, output of beam search layer. 
        scores (Variable): The associated scores of the ids, out put of beam
            search layer.
2070
        name (str): The name of this layer. It is optional.
D
dzhwinter 已提交
2071

2072
    Returns:
D
dzhwinter 已提交
2073 2074 2075 2076 2077
        tuple(Variable): a tuple of two output tensors: sentence_ids, sentence_scores.
        sentence_ids is a tensor with shape [size, length], where size is the
        beam size of beam search, and length is the length of each sentence. 
        Note that the length of sentences may vary.
        sentence_scores is a tensor with the same shape as sentence_ids.
D
dzhwinter 已提交
2078 2079 2080 2081 2082 2083 2084

    Examples:
        .. code-block:: python
            ids, scores = fluid.layers.beam_search(
                pre_ids, ids, scores, beam_size, end_id)
            sentence_ids, sentence_scores = fluid.layers.beam_search_decode(
                ids, scores)
2085
    """
Y
Yu Yang 已提交
2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105
    helper = LayerHelper('beam_search_decode', **locals())
    sentence_ids = helper.create_tmp_variable(dtype=ids.dtype)
    sentence_scores = helper.create_tmp_variable(dtype=ids.dtype)

    helper.append_op(
        type="beam_search_decode",
        inputs={"Ids": ids,
                "Scores": scores},
        outputs={
            "SentenceIds": sentence_ids,
            "SentenceScores": sentence_scores
        })

    return sentence_ids, sentence_scores


def conv2d_transpose(input,
                     num_filters,
                     output_size=None,
                     filter_size=None,
C
chengduoZH 已提交
2106 2107 2108
                     padding=0,
                     stride=1,
                     dilation=1,
2109
                     groups=None,
C
caoying03 已提交
2110
                     param_attr=None,
2111
                     bias_attr=None,
C
chengduoZH 已提交
2112
                     use_cudnn=True,
2113
                     act=None,
C
caoying03 已提交
2114
                     name=None):
Y
Yu Yang 已提交
2115
    """
2116 2117 2118 2119 2120 2121 2122 2123
    **Convlution2D transpose layer**

    The convolution2D transpose layer calculates the output based on the input,
    filter, and dilations, strides, paddings. Input(Input) and output(Output)
    are in NCHW format. Where N is batch size, C is the number of channels,
    H is the height of the feature, and W is the width of the feature.
    Parameters(dilations, strides, paddings) are two elements. These two elements
    represent height and width, respectively. The details of convolution transpose
2124 2125
    layer, please refer to the following explanation and references
    `therein <http://www.matthewzeiler.com/wp-content/uploads/2017/07/cvpr2010.pdf>`_.
2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137

    For each input :math:`X`, the equation is:

    .. math::

        Out = W \\ast X

    In the above equation:

    * :math:`X`: Input value, a tensor with NCHW format.
    * :math:`W`: Filter value, a tensor with MCHW format.
    * :math:`\\ast` : Convolution transpose operation.
2138 2139
    * :math:`Out`: Output value, the shape of :math:`Out` and :math:`X` may be
                   different.
Y
Yu Yang 已提交
2140

2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153
    Example:

        - Input:

          Input shape: $(N, C_{in}, H_{in}, W_{in})$

          Filter shape: $(C_{in}, C_{out}, H_f, W_f)$

        - Output:

          Output shape: $(N, C_{out}, H_{out}, W_{out})$

        Where
Y
Yu Yang 已提交
2154

2155 2156 2157 2158
        .. math::

           H_{out} &= (H_{in} - 1) * strides[0] - 2 * paddings[0] + dilations[0] * (H_f - 1) + 1 \\\\
           W_{out} &= (W_{in} - 1) * strides[1] - 2 * paddings[1] + dilations[1] * (W_f - 1) + 1
Y
Yu Yang 已提交
2159 2160

    Args:
2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193
        input(Variable): The input image with [N, C, H, W] format.
        num_filters(int): The number of the filter. It is as same as the output
            image channel.
        output_size(int|tuple|None): The output image size. If output size is a
            tuple, it must contain two integers, (image_H, image_W). This
            parameter only works when filter_size is None.
        filter_size(int|tuple|None): The filter size. If filter_size is a tuple,
            it must contain two integers, (filter_size_H, filter_size_W).
            Otherwise, the filter will be a square. None if use output size to
            calculate filter_size.
        padding(int|tuple): The padding size. If padding is a tuple, it must
            contain two integers, (padding_H, padding_W). Otherwise, the
            padding_H = padding_W = padding. Default: padding = 0.
        stride(int|tuple): The stride size. If stride is a tuple, it must
            contain two integers, (stride_H, stride_W). Otherwise, the
            stride_H = stride_W = stride. Default: stride = 1.
        dilation(int|tuple): The dilation size. If dilation is a tuple, it must
            contain two integers, (dilation_H, dilation_W). Otherwise, the
            dilation_H = dilation_W = dilation. Default: dilation = 1.
        groups(int): The groups number of the Conv2d transpose layer. Inspired by
            grouped convolution in Alex Krizhevsky's Deep CNN paper, in which
            when group=2, 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 connected to the second half of the input channels.
            Default: groups=1
        param_attr(ParamAttr): The parameters to the Conv2d_transpose Layer.
                               Default: None
        bias_attr(ParamAttr): Bias parameter for the Conv2d layer. Default: None
        use_cudnn(bool): Use cudnn kernel or not, it is valid only when the cudnn
            library is installed. Default: True
        act(str): Activation type. Default: None
        name(str|None): A name for this layer(optional). If set None, the layer
            will be named automatically.
Y
Yu Yang 已提交
2194 2195

    Returns:
2196
        Variable: The tensor variable storing the convolution transpose result.
2197 2198

    Raises:
2199 2200
        ValueError: If the shapes of input, filter_size, stride, padding and
                    groups mismatch.
2201 2202 2203 2204

    Examples:
       .. code-block:: python

2205 2206 2207 2208
          data = fluid.layers.data(
              name='data', shape=[3, 32, 32], dtype='float32')
          conv2d_transpose = fluid.layers.conv2d_transpose(
              input=data, num_filters=2, filter_size=3)
Y
Yu Yang 已提交
2209 2210 2211 2212 2213 2214
    """
    helper = LayerHelper("conv2d_transpose", **locals())
    if not isinstance(input, Variable):
        raise TypeError("Input of conv2d_transpose must be Variable")
    input_channel = input.shape[1]

C
chengduoZH 已提交
2215 2216 2217
    padding = utils.convert_to_list(padding, 2, 'padding')
    stride = utils.convert_to_list(stride, 2, 'stride')
    dilation = utils.convert_to_list(dilation, 2, 'dilation')
C
chengduoZH 已提交
2218

C
chengduoZH 已提交
2219 2220 2221
    if not isinstance(use_cudnn, bool):
        raise ValueError("use_cudnn should be True or False")

Y
Yu Yang 已提交
2222 2223 2224 2225 2226 2227 2228 2229
    if filter_size is None:
        if output_size is None:
            raise ValueError("output_size must be set when filter_size is None")
        if isinstance(output_size, int):
            output_size = [output_size, output_size]

        h_in = input.shape[2]
        w_in = input.shape[3]
C
chengduoZH 已提交
2230 2231 2232 2233 2234

        filter_size_h = (output_size[0] - (h_in - 1) * stride[0] + 2 *
                         padding[0] - 1) / dilation[0] + 1
        filter_size_w = (output_size[1] - (w_in - 1) * stride[1] + 2 *
                         padding[1] - 1) / dilation[1] + 1
Y
Yu Yang 已提交
2235
        filter_size = [filter_size_h, filter_size_w]
C
chengduoZH 已提交
2236 2237 2238
    else:
        filter_size = utils.convert_to_list(filter_size, 2,
                                            'conv2d_transpose.filter_size')
Y
Yu Yang 已提交
2239

2240 2241
    groups = 1 if groups is None else groups
    filter_shape = [input_channel, num_filters / groups] + filter_size
Y
Yu Yang 已提交
2242 2243 2244
    img_filter = helper.create_parameter(
        dtype=input.dtype, shape=filter_shape, attr=helper.param_attr)

2245
    pre_bias = helper.create_tmp_variable(dtype=input.dtype)
Y
Yu Yang 已提交
2246 2247 2248 2249
    helper.append_op(
        type='conv2d_transpose',
        inputs={'Input': [input],
                'Filter': [img_filter]},
2250
        outputs={'Output': pre_bias},
C
chengduoZH 已提交
2251 2252 2253 2254
        attrs={
            'strides': stride,
            'paddings': padding,
            'dilations': dilation,
2255
            'groups': groups,
C
chengduoZH 已提交
2256 2257
            'use_cudnn': use_cudnn
        })
Y
Yu Yang 已提交
2258

2259 2260
    pre_act = helper.append_bias_op(pre_bias, dim_start=1, dim_end=2)
    out = helper.append_activation(pre_act)
Y
Yu Yang 已提交
2261
    return out
Y
yangyaming 已提交
2262 2263


2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418
def conv3d_transpose(input,
                     num_filters,
                     output_size=None,
                     filter_size=None,
                     padding=0,
                     stride=1,
                     dilation=1,
                     groups=None,
                     param_attr=None,
                     bias_attr=None,
                     use_cudnn=True,
                     act=None,
                     name=None):
    """
    **Convlution3D transpose layer**

    The convolution3D transpose layer calculates the output based on the input,
    filter, and dilations, strides, paddings. Input(Input) and output(Output)
    are in NCDHW 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
    is the width of the feature. Parameters(dilations, strides, paddings) are
    two elements. These two elements represent height and width, respectively.
    The details of convolution transpose layer, please refer to the following
    explanation and references `therein <http://www.matthewzeiler.com/wp-content/uploads/2017/07/cvpr2010.pdf>`_.

    For each input :math:`X`, the equation is:

    .. math::

        Out = W \\ast X

    In the above equation:

    * :math:`X`: Input value, a tensor with NCDHW format.
    * :math:`W`: Filter value, a tensor with MCDHW format.
    * :math:`\\ast` : Convolution transpose operation.
    * :math:`Out`: Output value, the shape of :math:`Out` and :math:`X` may be
                   different.

    Example:

        - Input:

          Input shape: $(N, C_{in}, D_{in}, H_{in}, W_{in})$

          Filter shape: $(C_{in}, C_{out}, D_f, H_f, W_f)$

        - Output:

          Output shape: $(N, C_{out}, D_{out}, H_{out}, W_{out})$

        Where

        .. math::

           D_{out} &= (D_{in} - 1) * strides[0] - 2 * paddings[0] + dilations[0] * (D_f - 1) + 1 \\\\
           H_{out} &= (H_{in} - 1) * strides[1] - 2 * paddings[1] + dilations[1] * (H_f - 1) + 1 \\\\
           W_{out} &= (W_{in} - 1) * strides[2] - 2 * paddings[2] + dilations[2] * (W_f - 1) + 1

    Args:
        input(Variable): The input image with [N, C, D, H, W] format.
        num_filters(int): The number of the filter. It is as same as the output
            image channel.
        output_size(int|tuple|None): The output image size. If output size is a
            tuple, it must contain three integers, (image_D, image_H, image_W). This
            parameter only works when filter_size is None.
        filter_size(int|tuple|None): The filter size. If filter_size is a tuple,
            it must contain three integers, (filter_size_D, filter_size_H, filter_size_W).
            Otherwise, the filter will be a square. None if use output size to
            calculate filter_size.
        padding(int|tuple): The padding size. If padding is a tuple, it must
            contain three integers, (padding_D, padding_H, padding_W). Otherwise, the
            padding_D = padding_H = padding_W = padding. Default: padding = 0.
        stride(int|tuple): The stride size. If stride is a tuple, it must
            contain three integers, (stride_D, stride_H, stride_W). Otherwise, the
            stride_D = stride_H = stride_W = stride. Default: stride = 1.
        dilation(int|tuple): The dilation size. If dilation is a tuple, it must
            contain three integers, (dilation_D, dilation_H, dilation_W). Otherwise, the
            dilation_D = dilation_H = dilation_W = dilation. Default: dilation = 1.
        groups(int): The groups number of the Conv3d transpose layer. Inspired by
            grouped convolution in Alex Krizhevsky's Deep CNN paper, in which
            when group=2, 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 connected to the second half of the input channels.
            Default: groups=1
        param_attr(ParamAttr): The parameters to the Conv3d_transpose Layer.
            Default: None
        bias_attr(ParamAttr): Bias parameter for the Conv3d layer. Default: None
        use_cudnn(bool): Use cudnn kernel or not, it is valid only when the cudnn
            library is installed. Default: True
        act(str): Activation type. Default: None
        name(str|None): A name for this layer(optional). If set None, the layer
            will be named automatically.

    Returns:
        Variable: The tensor variable storing the convolution transpose result.

    Raises:
        ValueError: If the shapes of input, filter_size, stride, padding and
                    groups mismatch.

    Examples:
       .. code-block:: python

          data = fluid.layers.data(
              name='data', shape=[3, 12, 32, 32], dtype='float32')
          conv2d_transpose = fluid.layers.conv3d_transpose(
              input=data, num_filters=2, filter_size=3)
    """
    l_type = "conv3d_transpose"
    helper = LayerHelper(l_type, **locals())
    if not isinstance(input, Variable):
        raise TypeError("Input of conv3d_transpose must be Variable")
    input_channel = input.shape[1]

    padding = utils.convert_to_list(padding, 3, 'padding')
    stride = utils.convert_to_list(stride, 3, 'stride')
    dilation = utils.convert_to_list(dilation, 3, 'dilation')

    if not isinstance(use_cudnn, bool):
        raise ValueError("use_cudnn should be True or False")

    if filter_size is None:
        if output_size is None:
            raise ValueError("output_size must be set when filter_size is None")
        if isinstance(output_size, int):
            output_size = [output_size, output_size]

        d_in = input.shape[2]
        h_in = input.shape[3]
        w_in = input.shape[4]

        filter_size_d = (output_size[0] - (d_in - 1) * stride[0] + 2 *
                         padding[0] - 1) / dilation[0] + 1
        filter_size_h = (output_size[1] - (h_in - 1) * stride[1] + 2 *
                         padding[1] - 1) / dilation[1] + 1
        filter_size_w = (output_size[2] - (w_in - 1) * stride[2] + 2 *
                         padding[2] - 1) / dilation[2] + 1
        filter_size = [filter_size_d, filter_size_h, filter_size_w]
    else:
        filter_size = utils.convert_to_list(filter_size, 3,
                                            'conv3d_transpose.filter_size')

    groups = 1 if groups is None else groups
    filter_shape = [input_channel, num_filters / groups] + filter_size
    img_filter = helper.create_parameter(
        dtype=input.dtype, shape=filter_shape, attr=helper.param_attr)

    pre_bias = helper.create_tmp_variable(dtype=input.dtype)
    helper.append_op(
        type=l_type,
        inputs={'Input': [input],
                'Filter': [img_filter]},
        outputs={'Output': pre_bias},
        attrs={
C
chengduoZH 已提交
2419 2420 2421
            'strides': stride,
            'paddings': padding,
            'dilations': dilation,
2422
            'groups': groups,
C
chengduoZH 已提交
2423 2424
            'use_cudnn': use_cudnn
        })
Y
Yu Yang 已提交
2425

2426 2427
    pre_act = helper.append_bias_op(pre_bias, dim_start=1, dim_end=2)
    out = helper.append_activation(pre_act)
Y
Yu Yang 已提交
2428
    return out
Y
yangyaming 已提交
2429 2430


Y
yangyaming 已提交
2431
def sequence_expand(x, y, ref_level=-1, name=None):
2432
    """Sequence Expand Layer. This layer will expand the input variable **x**
Y
yangyaming 已提交
2433 2434 2435 2436
    according to specified level lod of **y**. Please note that lod level of
    **x** is at most 1 and rank of **x** is at least 2. When rank of **x**
    is greater than 2, then it would be viewed as a 2-D tensor.
    Following examples will explain how sequence_expand works:
2437 2438 2439 2440 2441

    .. code-block:: text

        * Case 1
            x is a LoDTensor:
Y
yangyaming 已提交
2442 2443
                x.lod  = [[0,   2,        4]]
                x.data = [[a], [b], [c], [d]]
2444 2445 2446 2447 2448 2449
                x.dims = [4, 1]

            y is a LoDTensor:
                y.lod = [[0,    2,    4],
                         [0, 3, 6, 7, 8]]

Y
yangyaming 已提交
2450
            ref_level: 0
2451

Y
yangyaming 已提交
2452 2453 2454
            then output is a 1-level LoDTensor:
                out.lod =  [[0,   2,        4,        6,        8]]
                out.data = [[a], [b], [a], [b], [c], [d], [c], [d]]
2455 2456 2457 2458
                out.dims = [8, 1]

        * Case 2
            x is a Tensor:
Y
yangyaming 已提交
2459
                x.data = [[a], [b], [c]]
2460 2461 2462
                x.dims = [3, 1]

            y is a LoDTensor:
Y
yangyaming 已提交
2463
                y.lod = [[0, 2, 2, 5]]
2464

Y
yangyaming 已提交
2465
            ref_level: -1
2466

Y
yangyaming 已提交
2467 2468 2469
            then output is a Tensor:
                out.data = [[a], [a], [c], [c], [c]]
                out.dims = [5, 1]
2470 2471 2472
    Args:
        x (Variable): The input variable which is a Tensor or LoDTensor.
        y (Variable): The input variable which is a LoDTensor.
Y
yangyaming 已提交
2473 2474
        ref_level (int): Lod level of `y` to be referred by `x`. If set to -1,
                         refer the last level of lod.
C
caoying03 已提交
2475
        name(str|None): A name for this layer(optional). If set None, the layer
Y
yangyaming 已提交
2476
                        will be named automatically.
2477 2478 2479 2480 2481 2482 2483 2484 2485 2486

    Returns:
        Variable: The expanded variable which is a LoDTensor.

    Examples:
        .. code-block:: python

            x = fluid.layers.data(name='x', shape=[10], dtype='float32')
            y = fluid.layers.data(name='y', shape=[10, 20],
                             dtype='float32', lod_level=1)
Y
yangyaming 已提交
2487
            out = layers.sequence_expand(x=x, y=y, ref_level=0)
2488
    """
Y
yangyaming 已提交
2489
    helper = LayerHelper('sequence_expand', input=x, **locals())
2490 2491 2492
    dtype = helper.input_dtype()
    tmp = helper.create_tmp_variable(dtype)
    helper.append_op(
Y
yangyaming 已提交
2493 2494 2495 2496 2497
        type='sequence_expand',
        inputs={'X': x,
                'Y': y},
        outputs={'Out': tmp},
        attrs={'ref_level': ref_level})
2498
    return tmp
2499 2500


Q
Qiao Longfei 已提交
2501 2502 2503
def beam_search(pre_ids, ids, scores, beam_size, end_id, level=0):
    '''
    This function implements the beam search algorithm.
2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514

    Args:
        pre_ids (Variable): ${pre_ids_comment}
        ids (Variable): ${ids_comment}
        scores (Variable): ${scores_comment}
        beam_size (int): ${beam_size_comment}
        end_id (int): ${end_id_comment}
        level (int): ${level_comment}
    
    Returns:
        tuple: a tuple of beam_search output variables: selected_ids, selected_scores
Q
Qiao Longfei 已提交
2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543
    '''
    helper = LayerHelper('beam_search', **locals())
    score_type = scores.dtype
    id_type = ids.dtype

    selected_scores = helper.create_tmp_variable(dtype=score_type)
    selected_ids = helper.create_tmp_variable(dtype=id_type)

    helper.append_op(
        type='beam_search',
        inputs={
            'pre_ids': pre_ids,
            'ids': ids,
            'scores': scores,
        },
        outputs={
            'selected_ids': selected_ids,
            'selected_scores': selected_scores,
        },
        attrs={
            # TODO(ChunweiYan) to assure other value support
            'level': level,
            'beam_size': beam_size,
            'end_id': end_id,
        })

    return selected_ids, selected_scores


Y
yangyaming 已提交
2544 2545 2546 2547
def lstm_unit(x_t,
              hidden_t_prev,
              cell_t_prev,
              forget_bias=0.0,
Y
yangyaming 已提交
2548
              param_attr=None,
C
caoying03 已提交
2549 2550
              bias_attr=None,
              name=None):
Y
yangyaming 已提交
2551 2552 2553 2554
    """Lstm unit layer. The equation of a lstm step is:

        .. math::

2555
            i_t & = \sigma(W_{x_i}x_{t} + W_{h_i}h_{t-1} + b_i)
Y
yangyaming 已提交
2556

2557
            f_t & = \sigma(W_{x_f}x_{t} + W_{h_f}h_{t-1} + b_f)
Y
yangyaming 已提交
2558

2559
            c_t & = f_tc_{t-1} + i_t tanh (W_{x_c}x_t + W_{h_c}h_{t-1} + b_c)
Y
yangyaming 已提交
2560

2561
            o_t & = \sigma(W_{x_o}x_{t} + W_{h_o}h_{t-1} + b_o)
Y
yangyaming 已提交
2562 2563 2564

            h_t & = o_t tanh(c_t)

2565 2566 2567 2568 2569 2570
    The inputs of lstm unit include :math:`x_t`, :math:`h_{t-1}` and
    :math:`c_{t-1}`. The 2nd dimensions of :math:`h_{t-1}` and :math:`c_{t-1}`
    should be same. The implementation separates the linear transformation and
    non-linear transformation apart. Here, we take :math:`i_t` as an example.
    The linear transformation is applied by calling a `fc` layer and the
    equation is:
Y
yangyaming 已提交
2571 2572 2573

        .. math::

2574
            L_{i_t} = W_{x_i}x_{t} + W_{h_i}h_{t-1} + b_i
Y
yangyaming 已提交
2575 2576 2577 2578 2579 2580 2581 2582

    The non-linear transformation is applied by calling `lstm_unit_op` and the
    equation is:

        .. math::

            i_t = \sigma(L_{i_t})

Y
yangyaming 已提交
2583
    This layer has two outputs including :math:`h_t` and :math:`o_t`.
Y
yangyaming 已提交
2584 2585

    Args:
Y
yangyaming 已提交
2586 2587 2588 2589 2590 2591
        x_t (Variable): The input value of current step, a 2-D tensor with shape
            M x N, M for batch size and N for input size.
        hidden_t_prev (Variable): The hidden value of lstm unit, a 2-D tensor
            with shape M x S, M for batch size and S for size of lstm unit.
        cell_t_prev (Variable): The cell value of lstm unit, a 2-D tensor with
            shape M x S, M for batch size and S for size of lstm unit.
Y
yangyaming 已提交
2592
        forget_bias (float): The forget bias of lstm unit.
Y
yangyaming 已提交
2593 2594
        param_attr (ParamAttr): The attributes of parameter weights, used to set
            initializer, name etc.
Y
yangyaming 已提交
2595 2596
        bias_attr (ParamAttr): The attributes of bias weights, if not False,
            bias weights will be created and be set to default value.
C
caoying03 已提交
2597 2598
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
Y
yangyaming 已提交
2599 2600

    Returns:
Y
yangyaming 已提交
2601
        tuple: The hidden value and cell value of lstm unit.
Y
yangyaming 已提交
2602 2603

    Raises:
2604 2605 2606 2607
        ValueError: The ranks of **x_t**, **hidden_t_prev** and **cell_t_prev**
                    not be 2 or the 1st dimensions of **x_t**, **hidden_t_prev**
                    and **cell_t_prev** not be the same or the 2nd dimensions of
                    **hidden_t_prev** and **cell_t_prev** not be the same.
Y
yangyaming 已提交
2608 2609 2610 2611 2612 2613

    Examples:

        .. code-block:: python

             x_t = fluid.layers.fc(input=x_t_data, size=10)
2614
             prev_hidden = fluid.layers.fc(input=prev_hidden_data, size=30)
Y
yangyaming 已提交
2615
             prev_cell = fluid.layers.fc(input=prev_cell_data, size=30)
Y
yangyaming 已提交
2616
             hidden_value, cell_value = fluid.layers.lstm_unit(x_t=x_t,
Y
yangyaming 已提交
2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632
                                                    hidden_t_prev=prev_hidden,
                                                    cell_t_prev=prev_cell)
    """
    helper = LayerHelper('lstm_unit', **locals())

    if len(x_t.shape) != 2:
        raise ValueError("Rank of x_t must be 2.")

    if len(hidden_t_prev.shape) != 2:
        raise ValueError("Rank of hidden_t_prev must be 2.")

    if len(cell_t_prev.shape) != 2:
        raise ValueError("Rank of cell_t_prev must be 2.")

    if x_t.shape[0] != hidden_t_prev.shape[0] or x_t.shape[
            0] != cell_t_prev.shape[0]:
Y
yangyaming 已提交
2633
        raise ValueError("The 1st dimensions of x_t, hidden_t_prev and "
2634 2635 2636 2637
                         "cell_t_prev must be the same.")

    if hidden_t_prev.shape[1] != cell_t_prev.shape[1]:
        raise ValueError("The 2nd dimensions of hidden_t_prev and "
Y
yangyaming 已提交
2638 2639
                         "cell_t_prev must be the same.")

Y
yangyaming 已提交
2640 2641 2642
    if bias_attr is None:
        bias_attr = ParamAttr()

Y
yangyaming 已提交
2643
    size = cell_t_prev.shape[1]
2644
    concat_out = concat(input=[x_t, hidden_t_prev], axis=1)
Y
yangyaming 已提交
2645 2646
    fc_out = fc(input=concat_out,
                size=4 * size,
Y
yangyaming 已提交
2647
                param_attr=param_attr,
2648
                bias_attr=bias_attr)
Y
yangyaming 已提交
2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660
    dtype = x_t.dtype
    c = helper.create_tmp_variable(dtype)
    h = helper.create_tmp_variable(dtype)

    helper.append_op(
        type='lstm_unit',
        inputs={"X": fc_out,
                "C_prev": cell_t_prev},
        outputs={"C": c,
                 "H": h},
        attrs={"forget_bias": forget_bias})

Y
yangyaming 已提交
2661
    return h, c
G
guosheng 已提交
2662 2663


C
caoying03 已提交
2664
def reduce_sum(input, dim=None, keep_dim=False, name=None):
G
guosheng 已提交
2665
    """
Y
yangyaming 已提交
2666
    Computes the sum of tensor elements over the given dimension.
G
guosheng 已提交
2667 2668 2669

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
W
whs 已提交
2670
        dim (list|int|None): The dimensions along which the sum is performed. If
Y
yangyaming 已提交
2671 2672
            :attr:`None`, sum all elements of :attr:`input` and return a
            Tensor variable with a single element, otherwise must be in the
W
whs 已提交
2673 2674
            range :math:`[-rank(input), rank(input))`. If :math:`dim[i] < 0`,
            the dimension to reduce is :math:`rank + dim[i]`.
2675
        keep_dim (bool|False): Whether to reserve the reduced dimension in the
Y
yangyaming 已提交
2676
            output Tensor. The result tensor will have one fewer dimension
G
guosheng 已提交
2677
            than the :attr:`input` unless :attr:`keep_dim` is true.
C
caoying03 已提交
2678 2679
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
G
guosheng 已提交
2680 2681 2682

    Returns:
        Variable: The reduced Tensor variable.
F
fengjiayi 已提交
2683

G
guosheng 已提交
2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694
    Examples:
        .. code-block:: python

            # x is a Tensor variable with following elements:
            #    [[0.2, 0.3, 0.5, 0.9]
            #     [0.1, 0.2, 0.6, 0.7]]
            # Each example is followed by the correspending output tensor.
            fluid.layers.reduce_sum(x)  # [3.5]
            fluid.layers.reduce_sum(x, dim=0)  # [0.3, 0.5, 1.1, 1.6]
            fluid.layers.reduce_sum(x, dim=-1)  # [1.9, 1.6]
            fluid.layers.reduce_sum(x, dim=1, keep_dim=True)  # [[1.9], [1.6]]
W
whs 已提交
2695 2696 2697 2698 2699 2700 2701 2702

            # x is a Tensor variable with shape [2, 2, 2] and elements as below:
            #      [[[1, 2], [3, 4]],
            #      [[5, 6], [7, 8]]]
            # Each example is followed by the correspending output tensor.
            fluid.layers.reduce_sum(x, dim=[1, 2]) # [10, 26]
            fluid.layers.reduce_sum(x, dim=[0, 1]) # [16, 20]

G
guosheng 已提交
2703 2704 2705
    """
    helper = LayerHelper('reduce_sum', **locals())
    out = helper.create_tmp_variable(dtype=helper.input_dtype())
W
whs 已提交
2706 2707
    if dim is not None and not isinstance(dim, list):
        dim = [dim]
G
guosheng 已提交
2708 2709 2710 2711 2712
    helper.append_op(
        type='reduce_sum',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
W
whs 已提交
2713
            'dim': dim if dim != None else [0],
G
guosheng 已提交
2714 2715 2716 2717
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out
G
guosheng 已提交
2718 2719


C
caoying03 已提交
2720
def reduce_mean(input, dim=None, keep_dim=False, name=None):
G
guosheng 已提交
2721
    """
Y
yangyaming 已提交
2722
    Computes the mean of tensor elements over the given dimension.
G
guosheng 已提交
2723 2724 2725

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
W
whs 已提交
2726
        dim (list|int|None): The dimensions along which the mean is computed. If
Y
yangyaming 已提交
2727 2728 2729
            :attr:`None`, compute the mean over all elements of :attr:`input`
            and return a Tensor variable with a single element, otherwise
            must be in the range :math:`[-rank(input), rank(input))`. If
W
whs 已提交
2730
            :math:`dim[i] < 0`, the dimension to reduce is :math:`rank + dim[i]`.
Y
yangyaming 已提交
2731 2732
        keep_dim (bool): Whether to reserve the reduced dimension in the
            output Tensor. The result tensor will have one fewer dimension
G
guosheng 已提交
2733
            than the :attr:`input` unless :attr:`keep_dim` is true.
C
caoying03 已提交
2734 2735
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
G
guosheng 已提交
2736 2737 2738

    Returns:
        Variable: The reduced Tensor variable.
F
fengjiayi 已提交
2739

G
guosheng 已提交
2740 2741 2742 2743 2744 2745 2746 2747 2748 2749
    Examples:
        .. code-block:: python

            # x is a Tensor variable with following elements:
            #    [[0.2, 0.3, 0.5, 0.9]
            #     [0.1, 0.2, 0.6, 0.7]]
            # Each example is followed by the correspending output tensor.
            fluid.layers.reduce_mean(x)  # [0.4375]
            fluid.layers.reduce_mean(x, dim=0)  # [0.15, 0.25, 0.55, 0.8]
            fluid.layers.reduce_mean(x, dim=-1)  # [0.475, 0.4]
F
stash  
fengjiayi 已提交
2750 2751
            fluid.layers.reduce_mean(
                x, dim=1, keep_dim=True)  # [[0.475], [0.4]]
W
whs 已提交
2752 2753 2754 2755 2756 2757 2758

            # x is a Tensor variable with shape [2, 2, 2] and elements as below:
            #      [[[1.0, 2.0], [3.0, 4.0]],
            #      [[5.0, 6.0], [7.0, 8.0]]]
            # Each example is followed by the correspending output tensor.
            fluid.layers.reduce_mean(x, dim=[1, 2]) # [2.5, 6.5]
            fluid.layers.reduce_mean(x, dim=[0, 1]) # [4.0, 5.0]
G
guosheng 已提交
2759 2760 2761
    """
    helper = LayerHelper('reduce_mean', **locals())
    out = helper.create_tmp_variable(dtype=helper.input_dtype())
W
whs 已提交
2762 2763
    if dim is not None and not isinstance(dim, list):
        dim = [dim]
G
guosheng 已提交
2764 2765 2766 2767 2768
    helper.append_op(
        type='reduce_mean',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
W
whs 已提交
2769
            'dim': dim if dim != None else [0],
G
guosheng 已提交
2770 2771 2772 2773
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out
2774 2775


C
caoying03 已提交
2776
def reduce_max(input, dim=None, keep_dim=False, name=None):
2777
    """
Y
yangyaming 已提交
2778
    Computes the maximum of tensor elements over the given dimension.
2779 2780 2781

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
W
whs 已提交
2782
        dim (list|int|None): The dimension along which the maximum is computed.
Y
yangyaming 已提交
2783 2784 2785
            If :attr:`None`, compute the maximum over all elements of
            :attr:`input` and return a Tensor variable with a single element,
            otherwise must be in the range :math:`[-rank(input), rank(input))`.
W
whs 已提交
2786
            If :math:`dim[i] < 0`, the dimension to reduce is :math:`rank + dim[i]`.
Y
yangyaming 已提交
2787 2788
        keep_dim (bool): Whether to reserve the reduced dimension in the
            output Tensor. The result tensor will have one fewer dimension
2789
            than the :attr:`input` unless :attr:`keep_dim` is true.
C
caoying03 已提交
2790 2791
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
2792 2793 2794

    Returns:
        Variable: The reduced Tensor variable.
Y
yangyaming 已提交
2795

2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806
    Examples:
        .. code-block:: python

            # x is a Tensor variable with following elements:
            #    [[0.2, 0.3, 0.5, 0.9]
            #     [0.1, 0.2, 0.6, 0.7]]
            # Each example is followed by the correspending output tensor.
            fluid.layers.reduce_max(x)  # [0.9]
            fluid.layers.reduce_max(x, dim=0)  # [0.2, 0.3, 0.6, 0.9]
            fluid.layers.reduce_max(x, dim=-1)  # [0.9, 0.7]
            fluid.layers.reduce_max(x, dim=1, keep_dim=True)  # [[0.9], [0.7]]
W
whs 已提交
2807 2808 2809 2810 2811 2812 2813

            # x is a Tensor variable with shape [2, 2, 2] and elements as below:
            #      [[[1.0, 2.0], [3.0, 4.0]],
            #      [[5.0, 6.0], [7.0, 8.0]]]
            # Each example is followed by the correspending output tensor.
            fluid.layers.reduce_max(x, dim=[1, 2]) # [4.0, 8.0]
            fluid.layers.reduce_max(x, dim=[0, 1]) # [7.0, 8.0]
2814 2815 2816
    """
    helper = LayerHelper('reduce_max', **locals())
    out = helper.create_tmp_variable(dtype=helper.input_dtype())
W
whs 已提交
2817 2818
    if dim is not None and not isinstance(dim, list):
        dim = [dim]
2819 2820 2821 2822 2823
    helper.append_op(
        type='reduce_max',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
W
whs 已提交
2824
            'dim': dim if dim != None else [0],
2825 2826 2827 2828 2829 2830
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out


C
caoying03 已提交
2831
def reduce_min(input, dim=None, keep_dim=False, name=None):
2832
    """
Y
yangyaming 已提交
2833
    Computes the minimum of tensor elements over the given dimension.
2834 2835 2836

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
W
whs 已提交
2837
        dim (list|int|None): The dimensions along which the minimum is computed.
Y
yangyaming 已提交
2838 2839 2840
            If :attr:`None`, compute the minimum over all elements of
            :attr:`input` and return a Tensor variable with a single element,
            otherwise must be in the range :math:`[-rank(input), rank(input))`.
W
whs 已提交
2841
            If :math:`dim[i] < 0`, the dimension to reduce is :math:`rank + dim[i]`.
Y
yangyaming 已提交
2842 2843
        keep_dim (bool): Whether to reserve the reduced dimension in the
            output Tensor. The result tensor will have one fewer dimension
2844
            than the :attr:`input` unless :attr:`keep_dim` is true.
C
caoying03 已提交
2845 2846
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
2847 2848 2849

    Returns:
        Variable: The reduced Tensor variable.
Y
yangyaming 已提交
2850

2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861
    Examples:
        .. code-block:: python

            # x is a Tensor variable with following elements:
            #    [[0.2, 0.3, 0.5, 0.9]
            #     [0.1, 0.2, 0.6, 0.7]]
            # Each example is followed by the correspending output tensor.
            fluid.layers.reduce_min(x)  # [0.1]
            fluid.layers.reduce_min(x, dim=0)  # [0.1, 0.2, 0.5, 0.7]
            fluid.layers.reduce_min(x, dim=-1)  # [0.2, 0.1]
            fluid.layers.reduce_min(x, dim=1, keep_dim=True)  # [[0.2], [0.1]]
W
whs 已提交
2862 2863 2864 2865 2866 2867 2868

            # x is a Tensor variable with shape [2, 2, 2] and elements as below:
            #      [[[1.0, 2.0], [3.0, 4.0]],
            #      [[5.0, 6.0], [7.0, 8.0]]]
            # Each example is followed by the correspending output tensor.
            fluid.layers.reduce_min(x, dim=[1, 2]) # [1.0, 5.0]
            fluid.layers.reduce_min(x, dim=[0, 1]) # [1.0, 2.0]
2869 2870 2871
    """
    helper = LayerHelper('reduce_min', **locals())
    out = helper.create_tmp_variable(dtype=helper.input_dtype())
W
whs 已提交
2872 2873
    if dim is not None and not isinstance(dim, list):
        dim = [dim]
2874 2875 2876 2877 2878
    helper.append_op(
        type='reduce_min',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
W
whs 已提交
2879
            'dim': dim if dim != None else [0],
2880 2881 2882 2883
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out
G
guosheng 已提交
2884 2885


2886 2887 2888 2889 2890 2891
def reduce_prod(input, dim=None, keep_dim=False, name=None):
    """
    Computes the product of tensor elements over the given dimension.

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
W
whs 已提交
2892
        dim (list|int|None): The dimensions along which the product is performed. If
2893 2894
            :attr:`None`, multipy all elements of :attr:`input` and return a
            Tensor variable with a single element, otherwise must be in the
W
whs 已提交
2895 2896
            range :math:`[-rank(input), rank(input))`. If :math:`dim[i] < 0`,
            the dimension to reduce is :math:`rank + dim[i]`.
2897 2898 2899
        keep_dim (bool|False): Whether to reserve the reduced dimension in the
            output Tensor. The result tensor will have one fewer dimension
            than the :attr:`input` unless :attr:`keep_dim` is true.
Y
yangyaming 已提交
2900
        name(str|None): A name for this layer(optional). If set None, the
Z
zhouhanqing 已提交
2901
            layer will be named automatically.
2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915

    Returns:
        Variable: The reduced Tensor variable.

    Examples:
        .. code-block:: python

            # x is a Tensor variable with following elements:
            #    [[0.2, 0.3, 0.5, 0.9]
            #     [0.1, 0.2, 0.6, 0.7]]
            # Each example is followed by the correspending output tensor.
            fluid.layers.reduce_prod(x)  # [0.0002268]
            fluid.layers.reduce_prod(x, dim=0)  # [0.02, 0.06, 0.3, 0.63]
            fluid.layers.reduce_prod(x, dim=-1)  # [0.027, 0.0084]
Y
yangyaming 已提交
2916
            fluid.layers.reduce_prod(x, dim=1,
Z
zhouhanqing 已提交
2917
                                     keep_dim=True)  # [[0.027], [0.0084]]
W
whs 已提交
2918 2919 2920 2921 2922 2923 2924

            # x is a Tensor variable with shape [2, 2, 2] and elements as below:
            #      [[[1.0, 2.0], [3.0, 4.0]],
            #      [[5.0, 6.0], [7.0, 8.0]]]
            # Each example is followed by the correspending output tensor.
            fluid.layers.reduce_prod(x, dim=[1, 2]) # [24.0, 1680.0]
            fluid.layers.reduce_prod(x, dim=[0, 1]) # [105.0, 384.0]
2925 2926 2927
    """
    helper = LayerHelper('reduce_prod', **locals())
    out = helper.create_tmp_variable(dtype=helper.input_dtype())
W
whs 已提交
2928 2929
    if dim is not None and not isinstance(dim, list):
        dim = [dim]
2930 2931 2932 2933 2934
    helper.append_op(
        type='reduce_prod',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
W
whs 已提交
2935
            'dim': dim if dim != None else [0],
2936 2937 2938 2939 2940 2941
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out


C
caoying03 已提交
2942
def split(input, num_or_sections, dim=-1, name=None):
G
guosheng 已提交
2943
    """
C
caoying03 已提交
2944
    Split the input tensor into multiple sub-tensors.
G
guosheng 已提交
2945 2946 2947

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
C
caoying03 已提交
2948 2949 2950 2951 2952
        num_or_sections (int|list): If :attr:`num_or_sections` is an integer,
            then the integer indicates the number of equal sized sub-tensors
            that the tensor will be divided into. If :attr:`num_or_sections`
            is a list of integers, the length of list indicates the number of
            sub-tensors and the integers indicate the sizes of sub-tensors'
G
guosheng 已提交
2953
            :attr:`dim` dimension orderly.
C
caoying03 已提交
2954
        dim (int): The dimension along which to split. If :math:`dim < 0`, the
G
guosheng 已提交
2955
            dimension to split along is :math:`rank(input) + dim`.
C
caoying03 已提交
2956 2957
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
G
guosheng 已提交
2958 2959

    Returns:
D
dzhwinter 已提交
2960
        list(Variable): The list of segmented tensor variables.
G
guosheng 已提交
2961 2962 2963 2964 2965 2966 2967 2968 2969

    Examples:
        .. code-block:: python

            # x is a Tensor variable with shape [3, 9, 5]:
            x0, x1, x2 = fluid.layers.split(x, num_or_sections=3, dim=1)
            x0.shape  # [3, 3, 5]
            x1.shape  # [3, 3, 5]
            x2.shape  # [3, 3, 5]
F
stash  
fengjiayi 已提交
2970 2971
            x0, x1, x2 = fluid.layers.split(
                x, num_or_sections=[2, 3, 4], dim=1)
G
guosheng 已提交
2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000
            x0.shape  # [3, 2, 5]
            x1.shape  # [3, 3, 5]
            x2.shape  # [3, 4, 5]
    """
    helper = LayerHelper('split', **locals())
    input_shape = input.shape
    dim = (len(input_shape) + dim) if dim < 0 else dim
    if isinstance(num_or_sections, int):
        assert num_or_sections > 1, 'num_or_sections must be more than 1.'
        num = num_or_sections
    else:
        assert len(num_or_sections) < input_shape[
            dim], 'len(num_or_sections) must not be more than input.shape[dim].'
        num = len(num_or_sections)
    outs = [
        helper.create_tmp_variable(dtype=helper.input_dtype())
        for i in range(num)
    ]
    helper.append_op(
        type='split',
        inputs={'X': input},
        outputs={'Out': outs},
        attrs={
            'num': num_or_sections if isinstance(num_or_sections, int) else 0,
            'sections': num_or_sections
            if isinstance(num_or_sections, list) else [],
            'axis': dim
        })
    return outs
C
caoying03 已提交
3001 3002 3003 3004 3005 3006 3007 3008 3009


def l2_normalize(x, axis, epsilon=1e-12, name=None):
    """
    **L2 normalize Layer**

    The l2 normalize layer normalizes `x` along dimension `axis` using an L2
    norm. For a 1-D tensor (`dim` is fixed to 0), this layer computes

3010 3011
    .. math::
    y = \frac{x}{ \sqrt{\sum {x^2} + epsion }}
C
caoying03 已提交
3012 3013 3014 3015 3016

    For `x` with more dimensions, this layer independently normalizes each 1-D
    slice along dimension `axis`.

    Args:
3017 3018 3019 3020 3021 3022 3023 3024
        x(Variable|list): The input tensor to l2_normalize layer.
        axis(int): The axis on which to apply normalization. If `axis < 0`,
            the dimension to normalization is rank(X) + axis. -1 is the
            last dimension.
        epsilon(float): The epsilon value is used to avoid division by zero,
            the defalut value is 1e-10.
        name(str|None): A name for this layer(optional). If set None, the layer
            will be named automatically.
C
caoying03 已提交
3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035


    Returns:
        Variable: The output tensor variable.

    Examples:
        .. code-block:: python

          data = fluid.layers.data(name="data",
                                   shape=(3, 17, 13),
                                   dtype="float32")
Y
ying 已提交
3036
          normed = fluid.layers.l2_normalize(x=data, axis=1)
C
caoying03 已提交
3037 3038
    """

F
fengjiayi 已提交
3039 3040
    if len(x.shape) == 1:
        axis = 0
C
caoying03 已提交
3041 3042
    helper = LayerHelper("l2_normalize", **locals())

3043 3044
    out = helper.create_tmp_variable(dtype=x.dtype)
    norm = helper.create_tmp_variable(dtype=x.dtype)
C
caoying03 已提交
3045
    helper.append_op(
3046 3047 3048 3049
        type="norm",
        inputs={"X": x},
        outputs={"Out": out,
                 "Norm": norm},
C
caoying03 已提交
3050
        attrs={
3051 3052
            "axis": 1 if axis is None else axis,
            "epsilon": epsilon,
C
caoying03 已提交
3053 3054
        })
    return out
3055 3056


3057
def matmul(x, y, transpose_x=False, transpose_y=False, name=None):
G
guosheng 已提交
3058
    """
Y
ying 已提交
3059 3060 3061 3062
    Applies matrix multiplication to two tensors.

    Currently, the input tensors' rank can be any, but when the rank of any
    inputs is bigger than 3, this two inputs' rank should be equal.
G
guosheng 已提交
3063

C
chengduoZH 已提交
3064
    The actual behavior depends on the shapes of :math:`x`, :math:`y` and the
3065
    flag values of :attr:`transpose_x`, :attr:`transpose_y`. Specifically:
G
guosheng 已提交
3066

3067 3068 3069 3070 3071
    - If a transpose flag is specified, the last two dimensions of the tensor
      are transposed. If the tensor is rank-1 of shape :math:`[D]`, then for
      :math:`x` it is treated as :math:`[1, D]` in nontransposed form and as
      :math:`[D, 1]` in transposed form, whereas for :math:`y` it is the
      opposite: It is treated as :math:`[D, 1]` in nontransposed form and as
3072
      :math:`[1, D]` in transposed form.
G
guosheng 已提交
3073

C
chengduoZH 已提交
3074
    - After transpose, the two tensors are 2-D or n-D and matrix multiplication
3075
      performs in the following way.
G
guosheng 已提交
3076

3077
      - If both are 2-D, they are multiplied like conventional matrices.
C
chengduoZH 已提交
3078
      - If either is n-D, it is treated as a stack of matrices residing in the
Y
ying 已提交
3079
        last two dimensions and a batched matrix multiply supporting broadcast
3080
        applies on the two tensors.
G
guosheng 已提交
3081

Y
ying 已提交
3082 3083
    Also note that if the raw tensor :math:`x` or :math:`y` is rank-1 and
    nontransposed, the prepended or appended dimension :math:`1` will be
C
chengduoZH 已提交
3084
    removed after matrix multiplication.
G
guosheng 已提交
3085 3086 3087

    Args:
        x (Variable): The input variable which is a Tensor or LoDTensor.
3088 3089 3090
        y (Variable): The input variable which is a Tensor or LoDTensor.
        transpose_x (bool): Whether to transpose :math:`x` before multiplication.
        transpose_y (bool): Whether to transpose :math:`y` before multiplication.
3091
        name(str|None): A name for this layer(optional). If set None, the layer
3092
            will be named automatically.
G
guosheng 已提交
3093 3094

    Returns:
3095
        Variable: The product Tensor variable.
G
guosheng 已提交
3096

G
guosheng 已提交
3097 3098 3099
    Examples:
        .. code-block:: python

3100
            # Examples to clarify shapes of the inputs and output
C
chengduoZH 已提交
3101 3102
            # x: [B, ..., M, K], y: [B, ..., K, N]
            fluid.layers.matmul(x, y)  # out: [B, ..., M, N]
Y
ying 已提交
3103

3104 3105
            # x: [B, M, K], y: [B, K, N]
            fluid.layers.matmul(x, y)  # out: [B, M, N]
Y
ying 已提交
3106

3107 3108
            # x: [B, M, K], y: [K, N]
            fluid.layers.matmul(x, y)  # out: [B, M, N]
Y
ying 已提交
3109

3110 3111
            # x: [M, K], y: [K, N]
            fluid.layers.matmul(x, y)  # out: [M, N]
Y
ying 已提交
3112 3113 3114 3115

            # x: [B, M, K], y: [K]
            fluid.layers.matmul(x, y)  # out: [B, M]

3116 3117
            # x: [K], y: [K]
            fluid.layers.matmul(x, y)  # out: [1]
3118

Y
ying 已提交
3119
            # x: [M], y: [N]
3120
            fluid.layers.matmul(x, y, True, True)  # out: [M, N]
G
guosheng 已提交
3121
    """
Y
ying 已提交
3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133

    def __check_input(x, y):
        if len(y.shape) > len(x.shape):
            raise ValueError(
                "Invalid inputs for matmul. "
                "x's rank should be always greater than or equal to y'rank.")

        x_shape = list(x.shape)
        y_shape = list(y.shape)
        if len(x_shape) == 1:
            x_shape = [1] + x_shape
        if len(y_shape) == 1:
Y
ying 已提交
3134
            y_shape = y_shape + [1]
Y
ying 已提交
3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150

        # check the inner 2 dimensions
        if transpose_x:
            x_shape[-2], x_shape[-1] = x_shape[-1], x_shape[-2]
        if transpose_y:
            y_shape[-2], y_shape[-1] = y_shape[-1], y_shape[-2]
        if x_shape[-1] != y_shape[-2]:
            raise ValueError("Invalid inputs for matmul.")

        if len(y_shape) > 2:
            for i, dim_x in enumerate(x_shape[:-2]):
                if dim_x != y_shape[i]:
                    raise ValueError("Invalid inputs for matmul.")

    __check_input(x, y)

3151
    helper = LayerHelper('matmul', **locals())
Y
ying 已提交
3152
    out = helper.create_tmp_variable(dtype=x.dtype)
G
guosheng 已提交
3153
    helper.append_op(
3154 3155 3156 3157 3158 3159 3160
        type='matmul',
        inputs={'X': x,
                'Y': y},
        outputs={'Out': out},
        attrs={'transpose_X': transpose_x,
               'transpose_Y': transpose_y})
    return out
3161 3162


3163
def topk(input, k, name=None):
Q
qingqing01 已提交
3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178
    """
    This operator is used to find values and indices of the k largest entries
    for the last dimension.

    If the input is a vector (rank=1), finds the k largest entries in the vector
    and outputs their values and indices as vectors. Thus values[j] is the j-th
    largest entry in input, and its index is indices[j].

    If the input is a Tensor with higher rank, this operator computes the top k
    entries along the last dimension.

    Args:
        input(Variable): The input variable which can be a vector or Tensor with
            higher rank.
        k(int): An integer value to specify the top k largest elements.
3179 3180
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
Q
qingqing01 已提交
3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211

    Returns:
        values(Variable): The k largest elements along each last dimensional
            slice.
        indices(Variable): The indices of values within the last dimension of
            input.

    Examples:
        .. code-block:: python

            top5_values, top5_indices = layers.topk(input, k=5)
    """
    shape = input.shape
    if k < 1 and k >= shape[-1]:
        raise ValueError("k must be greater than 0 and less than %d." %
                         (shape[-1]))

    helper = LayerHelper("top_k", **locals())
    values = helper.create_tmp_variable(dtype=input.dtype)
    indices = helper.create_tmp_variable(dtype="int64")
    helper.append_op(
        type="top_k",
        inputs={"X": [input]},
        outputs={"Out": [values],
                 "Indices": [indices]},
        attrs={"k": k})
    values.stop_gradient = True
    indices.stop_gradient = True
    return values, indices


W
wanghaoshuang 已提交
3212
def edit_distance(input, label, normalized=True, ignored_tokens=None,
W
wanghaoshuang 已提交
3213
                  name=None):
3214
    """
Y
ying 已提交
3215 3216 3217 3218 3219 3220 3221 3222 3223
    EditDistance operator computes the edit distances between a batch of
    hypothesis strings and their references. Edit distance, also called
    Levenshtein distance, measures how dissimilar two strings are by counting
    the minimum number of operations to transform one string into anthor.
    Here the operations include insertion, deletion, and substitution.

    For example, given hypothesis string A = "kitten" and reference
    B = "sitting", the edit distance is 3 for A will be transformed into B
    at least after two substitutions and one insertion:
W
wanghaoshuang 已提交
3224

Y
ying 已提交
3225
    "kitten" -> "sitten" -> "sittin" -> "sitting"
W
wanghaoshuang 已提交
3226

Y
ying 已提交
3227 3228 3229 3230
    Input(Hyps) is a LoDTensor consisting of all the hypothesis strings with
    the total number denoted by `batch_size`, and the separation is specified
    by the LoD information. And the `batch_size` reference strings are arranged
    in order in the same way in the LoDTensor Input(Refs).
W
wanghaoshuang 已提交
3231

Y
ying 已提交
3232 3233 3234
    Output(Out) contains the `batch_size` results and each stands for the edit
    distance for a pair of strings respectively. If Attr(normalized) is true,
    the edit distance will be divided by the length of reference string.
W
wanghaoshuang 已提交
3235

3236 3237 3238
    Args:
        input(Variable): The indices for hypothesis strings.
        label(Variable): The indices for reference strings.
Y
ying 已提交
3239 3240 3241 3242
        normalized(bool): Indicated whether to normalize the edit distance by
                          the length of reference string.
        ignored_tokens(list of int): Tokens that should be removed before
                                     calculating edit distance.
3243
        name (str): The name of this layer. It is optional.
3244

W
wanghaoshuang 已提交
3245
    Returns:
W
wanghaoshuang 已提交
3246
        Variable: sequence-to-sequence edit distance in shape [batch_size, 1].
W
wanghaoshuang 已提交
3247 3248 3249 3250 3251

    Examples:
        .. code-block:: python

            x = fluid.layers.data(name='x', shape=[8], dtype='float32')
3252 3253
            y = fluid.layers.data(name='y', shape=[7], dtype='float32')

3254
            cost = fluid.layers.edit_distance(input=x,label=y)
3255
    """
3256
    helper = LayerHelper("edit_distance", **locals())
3257

3258
    # remove some tokens from input and labels
W
wanghaoshuang 已提交
3259
    if ignored_tokens is not None and len(ignored_tokens) > 0:
3260 3261 3262 3263 3264 3265 3266
        erased_input = helper.create_tmp_variable(dtype="int64")
        erased_label = helper.create_tmp_variable(dtype="int64")

        helper.append_op(
            type="sequence_erase",
            inputs={"X": [input]},
            outputs={"Out": [erased_input]},
W
wanghaoshuang 已提交
3267
            attrs={"tokens": ignored_tokens})
3268 3269 3270 3271 3272
        input = erased_input

        helper.append_op(
            type="sequence_erase",
            inputs={"X": [label]},
W
whs 已提交
3273
            outputs={"Out": [erased_label]},
W
wanghaoshuang 已提交
3274
            attrs={"tokens": ignored_tokens})
3275 3276
        label = erased_label

3277 3278
    # edit distance op
    edit_distance_out = helper.create_tmp_variable(dtype="int64")
3279
    sequence_num = helper.create_tmp_variable(dtype="int64")
3280 3281 3282 3283
    helper.append_op(
        type="edit_distance",
        inputs={"Hyps": [input],
                "Refs": [label]},
3284 3285
        outputs={"Out": [edit_distance_out],
                 "SequenceNum": [sequence_num]},
3286 3287
        attrs={"normalized": normalized})

3288
    return edit_distance_out, sequence_num
3289 3290 3291 3292 3293


def ctc_greedy_decoder(input, blank, name=None):
    """
    This op is used to decode sequences by greedy policy by below steps:
Y
ying 已提交
3294 3295 3296 3297
    1. Get the indexes of max value for each row in input. a.k.a.
       numpy.argmax(input, axis=0).
    2. For each sequence in result of step1, merge repeated tokens between two
       blanks and delete all blanks.
3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326

    A simple example as below:

    .. code-block:: text

        Given:

        input.data = [[0.6, 0.1, 0.3, 0.1],
                      [0.3, 0.2, 0.4, 0.1],
                      [0.1, 0.5, 0.1, 0.3],
                      [0.5, 0.1, 0.3, 0.1],

                      [0.5, 0.1, 0.3, 0.1],
                      [0.2, 0.2, 0.2, 0.4],
                      [0.2, 0.2, 0.1, 0.5],
                      [0.5, 0.1, 0.3, 0.1]]

        input.lod = [[0, 4, 8]]

        Then:

        output.data = [[2],
                       [1],
                       [3]]

        output.lod = [[0, 2, 3]]

    Args:

Y
ying 已提交
3327 3328 3329 3330 3331 3332 3333 3334 3335
        input(Variable): (LoDTensor<float>), the probabilities of
                         variable-length sequences, which is a 2-D Tensor with
                         LoD information. It's shape is [Lp, num_classes + 1],
                         where Lp is the sum of all input sequences' length and
                         num_classes is the true number of classes. (not
                         including the blank label).
        blank(int): the blank label index of Connectionist Temporal
                    Classification (CTC) loss, which is in thehalf-opened
                    interval [0, num_classes + 1).
3336
        name (str): The name of this layer. It is optional.
3337 3338

    Returns:
3339
        Variable: CTC greedy decode result. If all the sequences in result were
3340
        empty, the result LoDTensor will be [-1] with LoD [[0]] and dims [1, 1].
3341 3342 3343 3344 3345

    Examples:
        .. code-block:: python

            x = fluid.layers.data(name='x', shape=[8], dtype='float32')
W
wanghaoshuang 已提交
3346

3347
            cost = fluid.layers.ctc_greedy_decoder(input=x, blank=0)
W
wanghaoshuang 已提交
3348
    """
3349
    helper = LayerHelper("ctc_greedy_decoder", **locals())
Q
qingqing01 已提交
3350
    _, topk_indices = topk(input, k=1)
3351 3352 3353 3354 3355 3356

    # ctc align op
    ctc_out = helper.create_tmp_variable(dtype="int64")
    helper.append_op(
        type="ctc_align",
        inputs={"Input": [topk_indices]},
W
wanghaoshuang 已提交
3357
        outputs={"Output": [ctc_out]},
3358 3359
        attrs={"merge_repeated": True,
               "blank": blank})
3360
    return ctc_out
3361 3362


F
fengjiayi 已提交
3363
def warpctc(input, label, blank=0, norm_by_times=False):
W
wanghaoshuang 已提交
3364
    """
3365 3366
    An operator integrating the open source Warp-CTC library
    (https://github.com/baidu-research/warp-ctc)
W
wanghaoshuang 已提交
3367
    to compute Connectionist Temporal Classification (CTC) loss.
3368 3369
    It can be aliased as softmax with CTC, since a native softmax activation is
    interated to the Warp-CTC library, to to normlize values for each row of the
W
wanghaoshuang 已提交
3370 3371 3372
    input tensor.

    Args:
3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389
        input(Variable): (LodTensor, default: LoDTensor<float>),
            the unscaled probabilities of variable-length sequences,
            which is a 2-D Tensor with LoD information.
            It's shape is [Lp, num_classes + 1], where Lp is the sum of all input
            sequences' length and num_classes is the true number of classes.
            (not including the blank label).
        label(Variable): (LodTensor, default: LoDTensor<int>), the ground truth
            of variable-length sequence, which is a 2-D Tensor with LoD
            information. It is of the shape [Lg, 1], where Lg is th sum of
            all labels' length.
        blank (int): default 0, the blank label index of Connectionist
            Temporal Classification (CTC) loss, which is in the
            half-opened interval [0, num_classes + 1).
        norm_by_times (bool): default false, whether to normalize
            the gradients by the number of time-step, which is also the
            sequence's length. There is no need to normalize the gradients
            if warpctc layer was follewed by a mean_op.
W
wanghaoshuang 已提交
3390 3391

    Returns:
3392 3393
        Variable: The Connectionist Temporal Classification (CTC) loss,
        which is a 2-D Tensor of the shape [batch_size, 1].
W
wanghaoshuang 已提交
3394 3395 3396

    Examples:
        .. code-block:: python
3397 3398 3399 3400
            y = layers.data(
                name='y', shape=[11, 8], dtype='float32', lod_level=1)
            y_predict = layers.data(
                name='y_predict', shape=[11, 1], dtype='float32')
W
wanghaoshuang 已提交
3401 3402 3403
            cost = layers.warpctc(input=y_predict, label=y)

    """
F
fengjiayi 已提交
3404
    helper = LayerHelper('warpctc', **locals())
W
wanghaoshuang 已提交
3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415
    loss_out = helper.create_tmp_variable(dtype=input.dtype)
    grad_out = helper.create_tmp_variable(dtype=input.dtype)
    helper.append_op(
        type='warpctc',
        inputs={'Logits': [input],
                'Label': [label]},
        outputs={'WarpCTCGrad': [grad_out],
                 'Loss': [loss_out]},
        attrs={'blank': blank,
               'norm_by_times': norm_by_times})
    return loss_out
3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447


def sequence_reshape(input, new_dim):
    """
    **Sequence Reshape Layer**

    This layer will rearrange the input sequences. The new dimension is set by
    user. Length of each sequence is computed according to original length,
    original dimension and new dimension. The following example will help to
    illustrate the function of this layer:

    .. code-block:: text

        x is a LoDTensor:
            x.lod  = [[0, 2, 6]]
            x.data = [[1, 2], [3, 4],
                      [5, 6], [7, 8], [9, 10], [11, 12]]
            x.dims = [6, 2]

        set new_dim = 4

        then out is a LoDTensor:
            out.lod  = [[0, 1, 3]]
            out.data = [[1, 2, 3, 4],
                        [5, 6, 7, 8], [9, 10, 11, 12]]
            out.dims = [3, 4]

    Currently, only 1-level LoDTensor is supported and please make sure
    (original length * original dimension) can be divided by new dimension with
    no remainder for each sequence.

    Args:
3448 3449 3450
        input (Variable): (LodTensor, default: LoDTensor<float>), a 2-D LoDTensor
            with shape being [N, M] where M for dimension.
        new_dim (int): New dimension which the input LoDTensor is reshaped to.
3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469

    Returns:
        Variable: Reshaped LoDTensor according to new dimension.

    Examples:
        .. code-block:: python

            x = fluid.layers.data(name='x', shape=[5, 20],
                              dtype='float32', lod_level=1)
            x_reshaped = layers.sequence_reshape(input=x, new_dim=10)
    """
    helper = LayerHelper('sequence_reshape', **locals())
    out = helper.create_tmp_variable(helper.input_dtype())
    helper.append_op(
        type='sequence_reshape',
        inputs={'X': [input]},
        outputs={'Out': [out]},
        attrs={'new_dim': new_dim})
    return out
Y
ying 已提交
3470 3471


3472 3473 3474 3475
# FIXME(wuyi): let docstring_checker.py understand @autodoc.
# For now, the comments in c++ use types like Tensor, but in python side
# the type is often "Variable", and arguments may vary.
@templatedoc(op_type="nce")
Y
Yang Yu 已提交
3476 3477 3478 3479 3480 3481 3482
def nce(input,
        label,
        num_total_classes,
        sample_weight=None,
        param_attr=None,
        bias_attr=None,
        num_neg_samples=None):
3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497
    """
    ${comment}

    Args:
        input (Variable): input variable.
        label (Variable): label.
        num_total_classes (int):${num_total_classes_comment}
        sample_weight (int): ${sample_weight_comment}
        param_attr (ParamAttr|None): attributes for parameter
        bias_attr (ParamAttr|None): attributes for bias
        num_neg_samples (int): ${num_neg_samples_comment}
    
    Returns:
        Variable: output of nce layer.
    """
Y
Yang Yu 已提交
3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516
    helper = LayerHelper('nce', **locals())
    assert isinstance(input, Variable)
    dim = input.shape[1]
    assert isinstance(label, Variable)
    num_true_class = label.shape[1]
    w = helper.create_parameter(
        attr=helper.param_attr,
        shape=[num_total_classes, dim],
        is_bias=False,
        dtype=input.dtype)
    b = helper.create_parameter(
        attr=helper.bias_attr,
        shape=[num_total_classes, 1],
        is_bias=True,
        dtype=input.dtype)
    cost = helper.create_tmp_variable(dtype=input.dtype)
    sample_logits = helper.create_tmp_variable(dtype=input.dtype)
    sample_labels = helper.create_tmp_variable(dtype=label.dtype)

Y
Yang Yu 已提交
3517 3518 3519 3520 3521 3522 3523 3524 3525
    if num_neg_samples is None:
        num_neg_samples = 10
    else:
        num_neg_samples = int(num_neg_samples)

    attrs = {
        'num_total_classes': int(num_total_classes),
        'num_neg_samples': num_neg_samples
    }
Y
Yang Yu 已提交
3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541

    helper.append_op(
        type='nce',
        inputs={
            'Input': input,
            'Label': label,
            'Weight': w,
            'Bias': b,
            'SampleWeight': sample_weight if sample_weight is not None else []
        },
        outputs={
            'Cost': cost,
            'SampleLogits': sample_logits,
            'SampleLabels': sample_labels
        },
        attrs=attrs)
Y
Yang Yu 已提交
3542
    return cost / (num_neg_samples + 1)
3543 3544


Y
fix ci.  
ying 已提交
3545
def transpose(x, perm, name=None):
Y
ying 已提交
3546 3547 3548 3549 3550 3551 3552 3553 3554
    """
    **transpose Layer**

    Permute the dimensions of `input` according to `perm`.

    The `i`-th dimension  of the returned tensor will correspond to the
    perm[i]-th dimension of `input`.

    Args:
3555 3556 3557
        x (Variable): The input Tensor.
        perm (list): A permutation of the dimensions of `input`.
        name (str): The name of this layer. It is optional.
Y
ying 已提交
3558 3559 3560 3561 3562 3563 3564 3565

    Returns:
        Variable: A transposed Tensor.

    Examples:
        .. code-block:: python

            x = fluid.layers.data(name='x', shape=[5, 10, 15], dtype='float32')
Y
fix ci.  
ying 已提交
3566
            x_transposed = layers.transpose(x, perm=[1, 0, 2])
Y
ying 已提交
3567 3568
    """

Y
fix ci.  
ying 已提交
3569
    if len(perm) != len(x.shape):
Y
ying 已提交
3570 3571 3572
        raise ValueError(
            "Input(perm) is the permutation of dimensions of Input(input). "
            "It's length shoud be equal to Input(input)'s rank.")
Y
ying 已提交
3573 3574 3575 3576 3577 3578
    for idx, dim in enumerate(perm):
        if dim >= len(x.shape):
            raise ValueError(
                "Each element in perm should be less than x's rank. "
                "%d-th element in perm is %d which accesses x's rank %d." %
                (idx, perm[idx], len(x.shape)))
Y
ying 已提交
3579 3580

    helper = LayerHelper('transpose', **locals())
Y
fix ci.  
ying 已提交
3581
    out = helper.create_tmp_variable(x.dtype)
Y
ying 已提交
3582 3583
    helper.append_op(
        type='transpose',
Y
fix ci.  
ying 已提交
3584
        inputs={'X': [x]},
Y
ying 已提交
3585 3586 3587
        outputs={'Out': [out]},
        attrs={'axis': perm})
    return out
3588 3589


3590
def im2sequence(input, filter_size=1, stride=1, padding=0, name=None):
3591
    """
3592 3593 3594 3595 3596 3597 3598
    Extracts image patches from the input tensor to form a tensor of shape
    {input.batch_size * output_height * output_width, filter_size_H *
    filter_size_W * input.channels} which is similar with im2col.
    This op use filter / kernel to scan images and convert these images to
    sequences. After expanding, the number of time step are
    output_height * output_width for an image, in which output_height and
    output_width are calculated by below equation:
3599 3600 3601 3602 3603 3604 3605 3606 3607 3608

    .. math::

        output\_size = 1 + \
            (2 * padding + img\_size - block\_size + stride - 1) / stride

    And the dimension of each time step is block_y * block_x * input.channels.

    Args:
        input (Variable): The input should be a tensor in NCHW format.
W
wanghaoshuang 已提交
3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626

        filter_size(int|tuple|None): The filter size. If filter_size is a tuple,
            it must contain two integers, (filter_size_H, filter_size_W).
            Otherwise, the filter will be a square.

        stride(int|tuple): The stride size. If stride is a tuple, it must
            contain two integers, (stride_H, stride_W). Otherwise, the
            stride_H = stride_W = stride. Default: stride = 1.

        padding(int|tuple): The padding size. If padding is a tuple, it can
            contain two integers like (padding_H, padding_W) which means
            padding_up = padding_down = padding_H and
            padding_left = padding_right = padding_W. Or it can use
            (padding_up, padding_left, padding_down, padding_right) to indicate
            paddings of four direction. Otherwise, a scalar padding means
            padding_up = padding_down = padding_left = padding_right = padding
            Default: padding = 0.

3627 3628 3629
        name (int): The name of this layer. It is optional.

    Returns:
W
wanghaoshuang 已提交
3630 3631 3632 3633 3634
        output: The output is a LoDTensor with shape
        {input.batch_size * output_height * output_width,
        filter_size_H * filter_size_W * input.channels}.
        If we regard output as a matrix, each row of this matrix is
        a step of a sequence.
3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661

    Examples:

        .. code-block:: text

            Given:

            x = [[[[ 6.  2.  1.]
                   [ 8.  3.  5.]
                   [ 0.  2.  6.]]

                  [[ 2.  4.  4.]
                   [ 6.  3.  0.]
                   [ 6.  4.  7.]]]

                 [[[ 6.  7.  1.]
                   [ 5.  7.  9.]
                   [ 2.  4.  8.]]

                  [[ 1.  2.  1.]
                   [ 1.  3.  5.]
                   [ 9.  0.  8.]]]]

            x.dims = {2, 2, 3, 3}

            And:

W
wanghaoshuang 已提交
3662 3663 3664
            filter = [2, 2]
            stride = [1, 1]
            padding = [0, 0]
3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680

            Then:

            output.data = [[ 6.  2.  8.  3.  2.  4.  6.  3.]
                           [ 2.  1.  3.  5.  4.  4.  3.  0.]
                           [ 8.  3.  0.  2.  6.  3.  6.  4.]
                           [ 3.  5.  2.  6.  3.  0.  4.  7.]
                           [ 6.  7.  5.  7.  1.  2.  1.  3.]
                           [ 7.  1.  7.  9.  2.  1.  3.  5.]
                           [ 5.  7.  2.  4.  1.  3.  9.  0.]
                           [ 7.  9.  4.  8.  3.  5.  0.  8.]]

            output.dims = {8, 9}

            output.lod = [[0, 4, 8]]

D
dzhwinter 已提交
3681
     Examples:
3682 3683 3684

        .. code-block:: python

3685 3686
            output = fluid.layers.im2sequence(
                input=layer, stride=[1, 1], filter_size=[2, 2])
3687 3688

    """
W
wanghaoshuang 已提交
3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699

    if isinstance(filter_size, int):
        filter_size = [filter_size, filter_size]
    if isinstance(stride, int):
        stride = [stride, stride]
    if isinstance(padding, int):
        padding = [padding, padding]
    if len(padding) == 2:
        padding.append(padding[0])
        padding.append(padding[1])

3700
    helper = LayerHelper('im2sequence', **locals())
3701 3702
    out = helper.create_tmp_variable(dtype=helper.input_dtype())
    helper.append_op(
3703
        type='im2sequence',
3704 3705 3706
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
W
wanghaoshuang 已提交
3707 3708 3709
            'kernels': filter_size,
            'strides': stride,
            'paddings': padding,
3710 3711
        })
    return out
3712 3713


3714 3715 3716 3717
def row_conv(input, future_context_size, param_attr=None, act=None):
    """Row Conv Operator. This layer will apply lookahead convolution to
    **input**. The input variable should be a 2D LoDTensor with shape [T, D].
    Parameters with shape [future_context_size + 1, D] will be created. The math
Y
yangyaming 已提交
3718
    equation of row convolution is as follows:
3719 3720 3721 3722 3723 3724 3725

    .. math::
        Out_{i} = \sum_{j = i} ^ {i + \\tau} X_{j} \odot W_{i - j}

    In the above equation:

    * :math:`Out_{i}`: The i-th row of output variable with shape [1, D].
Y
yangyaming 已提交
3726
    * :math:`\\tau`: Future context size.
3727 3728 3729 3730 3731 3732 3733 3734 3735 3736
    * :math:`X_{j}`: The j-th row of input variable with shape [1, D].
    * :math:`W_{i-j}`: The (i-j)-th row of parameters with shape [1, D].

    More details about row_conv please refer to the paper \
    (http://www.cs.cmu.edu/~dyogatam/papers/wang+etal.iclrworkshop2016.pdf) and
    the design document \
    (https://github.com/PaddlePaddle/Paddle/issues/2228#issuecomment-303903645).

    Args:
        input (Variable): Input variable, a 2D LoDTensor with shape [T, D].
Y
yangyaming 已提交
3737 3738
        future_context_size (int): Future context size. Please note, the shape
            of convolution kernel is [future_context_size + 1, D].
3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763
        param_attr (ParamAttr): Attributes of parameters, including
            name, initializer etc.
        act (str): Non-linear activation to be applied to output variable.

    Returns:
        Variable: The output tensor with same shape as input tensor.

    Examples:
        .. code-block:: python

            x = fluid.layers.data(name='x', shape=[16],
                            dtype='float32', lod_level=1)
            out = fluid.layers.row_conv(input=x, future_context_size=2)
    """
    helper = LayerHelper('row_conv', **locals())
    dtype = helper.input_dtype()
    filter_shape = [future_context_size + 1, input.shape[1]]
    filter_param = helper.create_parameter(
        attr=helper.param_attr, shape=filter_shape, dtype=dtype)
    out = helper.create_tmp_variable(dtype)
    helper.append_op(
        type='row_conv',
        inputs={'X': [input],
                'Filter': [filter_param]},
        outputs={'Out': [out]})
Y
yangyaming 已提交
3764
    return helper.append_activation(out)
3765 3766


3767 3768 3769 3770
def multiplex(inputs, index):
    """
    **Multiplex Layer**

Y
yangyaming 已提交
3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785
    Referring to the given index variable, this layer selects rows from the
    input variables to construct a multiplex variable. Assuming that there are
    :math:`m` input variables and :math:`I_i` represents the i-th input
    variable and :math:`i` is in [0, :math:`m`). All input variables are
    tensors with same shape [:math:`d_0`, :math:`d_1`, ..., :math:`d_R`].
    Please note that rank of the input tensor should be at least 2. Each input
    variable will be treated as a 2-D matrix with shape [:math:`M`, :math:`N`]
    where :math:`M` for :math:`d_0` and :math:`N` for :math:`d_1` * :math:`d_2`
    * ... * :math:`d_R`. Let :math:`I_i[j]` be the j-th row of the i-th input
    variable. The given index variable should be a 2-D tensor with shape
    [:math:`M`, 1]. Let `ID[i]` be the i-th index value of the index variable.
    Then the output variable will be a tensor with shape [:math:`d_0`,
    :math:`d_1`, ..., :math:`d_R`]. If we treat the output tensor as a 2-D
    matrix with shape [:math:`M`, :math:`N`] and let :math:`O[i]` be the i-th
    row of the matrix, then `O[i]` is equal to :math:`I_{ID[i]}[i]`.
3786 3787

    Args:
3788
        inputs (list): A list of variables to gather from. All variables have the
Y
yangyaming 已提交
3789
                same shape and the rank is at least 2.
3790
        index (Variable): Tensor<int32>, index variable which is a 2-D tensor
Y
yangyaming 已提交
3791
                with shape [M, 1] where M is the batch size.
3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804

    Returns:
        Variable: Multiplex variable gathered from input variables.

    Examples:
        .. code-block:: python

            x1 = fluid.layers.data(name='x1', shape=[4], dtype='float32')
            x2 = fluid.layers.data(name='x2', shape=[4], dtype='float32')
            index = fluid.layers.data(name='index', shape=[1], dtype='int32')
            out = fluid.layers.multiplex(inputs=[x1, x2], index=index)
    """
    helper = LayerHelper('multiplex', **locals())
Y
yangyaming 已提交
3805 3806 3807 3808 3809 3810

    if not isinstance(inputs, list) and len(inputs) < 2:
        raise ValueError("inputs should be a list object and contains at least "
                         "2 elements.")

    out = helper.create_tmp_variable(inputs[0].dtype)
3811 3812 3813 3814 3815 3816
    helper.append_op(
        type='multiplex',
        inputs={'X': inputs,
                'Ids': index},
        outputs={'Out': [out]})
    return out
3817 3818 3819 3820 3821


def softmax_with_cross_entropy(logits, label, soft_label=False):
    """
    **Softmax With Cross Entropy Operator.**
3822

3823 3824 3825 3826
    Cross entropy loss with softmax is used as the output layer extensively. This
    operator computes the softmax normalized values for each row of the input
    tensor, after which cross-entropy loss is computed. This provides a more
    numerically stable gradient.
3827

3828 3829 3830
    Because this operator performs a softmax on logits internally, it expects
    unscaled logits. This operator should not be used with the output of
    softmax operator since that would produce incorrect results.
3831

3832 3833 3834
    When the attribute soft_label is set false, this operators expects mutually
    exclusive hard labels, each sample in a batch is in exactly one class with a
    probability of 1.0. Each sample in the batch will have a single label.
3835

3836
    The equation is as follows:
3837

3838
    1) Hard label (one-hot label, so every sample has exactly one class)
3839

3840 3841 3842 3843
    .. math::

        loss_j =  -\\text{logit}_{label_j} +
        \\log\\left(\\sum_{i=0}^{K}\\exp(\\text{logit}_i)\\right), j = 1,..., K
3844

3845 3846 3847
    2) Soft label (each sample can have a distribution over all classes)

    .. math::
3848

3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869
        loss_j =  -\\sum_{i=0}^{K}\\text{label}_i
        \\left(\\text{logit}_i - \\log\\left(\\sum_{i=0}^{K}
        \\exp(\\text{logit}_i)\\right)\\right), j = 1,...,K

    Args:
        logits (Variable): The unscaled log probabilities, which is a 2-D tensor
            with shape [N x K]. N is the batch_size, and K is the class number.
        label (Variable): The ground truth which is a 2-D tensor. If soft_label
            is set to false, Label is a Tensor<int64> with shape [N x 1]. If
            soft_label is set to true, Label is a Tensor<float/double> with
        soft_label (bool): A flag to indicate whether to interpretate the given
            labels as soft labels. By default, `soft_label` is set to False.
    Returns:
        Variable: The cross entropy loss is a 2-D tensor with shape [N x 1].

    Examples:
        .. code-block:: python

            data = fluid.layers.data(name='data', shape=[128], dtype='float32')
            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
            fc = fluid.layers.fc(input=data, size=100)
F
stash  
fengjiayi 已提交
3870 3871
            out = fluid.layers.softmax_with_cross_entropy(
                logits=fc, label=label)
3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889
    """
    helper = LayerHelper('softmax_with_cross_entropy', **locals())
    softmax = helper.create_tmp_variable(dtype=logits.dtype)
    loss = helper.create_tmp_variable(dtype=logits.dtype)
    helper.append_op(
        type='softmax_with_cross_entropy',
        inputs={'Logits': logits,
                'Label': label},
        outputs={'Softmax': softmax,
                 'Loss': loss},
        attrs={'soft_label': soft_label})
    return loss


def smooth_l1(x, y, inside_weight=None, outside_weight=None, sigma=None):
    """
    **Smooth L1 Loss Operator. **

Q
qingqing01 已提交
3890
    This operator computes the smooth L1 loss for X and Y.
3891
    The operator takes the first dimension of X and Y as batch size.
Q
qingqing01 已提交
3892
    For each instance, it computes the smooth L1 loss element by element first
3893
    and then sums all the losses. So the shape of Out is [batch_size, 1].
3894

3895 3896
    Args:
        x (Variable): A tensor with rank at least 2. The input value of smooth
Q
qingqing01 已提交
3897
            L1 loss op with shape [batch_size, dim1, ..., dimN].
3898
        y (Variable): A tensor with rank at least 2. The target value of smooth
Q
qingqing01 已提交
3899
            L1 loss op with same shape as x.
3900 3901 3902 3903 3904 3905
        inside_weight (Variable|None):  A tensor with rank at least 2. This
            input is optional and should have same shape with x. If provided,
            the result of (x - y) will be multiplied by this tensor element by
            element.
        outside_weight (Variable|None): A tensor with rank at least 2. This
            input is optional and should have same shape with x. If provided,
Q
qingqing01 已提交
3906
            the out smooth L1 loss will be multiplied by this tensor element
3907
            by element.
Q
qingqing01 已提交
3908
        sigma (float|None): Hyper parameter of smooth L1 loss op. A float scalar
3909 3910
            with default value 1.0.
    Returns:
Q
qingqing01 已提交
3911
        Variable: A tensor with rank be 2. The output smooth L1 loss with
3912 3913 3914 3915 3916 3917
            shape [batch_size, 1].

    Examples:
        .. code-block:: python

            data = fluid.layers.data(name='data', shape=[128], dtype='float32')
F
stash  
fengjiayi 已提交
3918 3919
            label = fluid.layers.data(
                name='label', shape=[100], dtype='float32')
3920
            fc = fluid.layers.fc(input=data, size=100)
F
fengjiayi 已提交
3921
            out = fluid.layers.smooth_l1(x=fc, y=label)
3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937
    """
    helper = LayerHelper('smooth_l1_loss', **locals())
    diff = helper.create_tmp_variable(dtype=x.dtype)
    loss = helper.create_tmp_variable(dtype=x.dtype)
    helper.append_op(
        type='smooth_l1_loss',
        inputs={
            'X': x,
            'Y': y,
            'InsideWeight': inside_weight,
            'OutsideWeight': outside_weight
        },
        outputs={'Diff': diff,
                 'Out': loss},
        attrs={'sigma': sigma})
    return loss
3938 3939 3940 3941 3942 3943 3944 3945 3946


def one_hot(input, depth):
    """
    One Hot Operator. This operator creates the one-hot representations for input
    index values. The following example will help to explain the function of this
    operator.

    Args:
F
fengjiayi 已提交
3947
        input(variable):  A Tensor/LodTensor of indices, last dimension must be 1.
3948 3949 3950 3951 3952 3953
        depth(scalar): an interger defining the depth of the one hot dimension.

    Returns:
         The one-hot tensor or LodTensor, same as input.

    Examples:
C
caoying03 已提交
3954 3955
        .. code-block:: python

3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976
        X is a LoDTensor:
          X.lod = [[0, 1, 4]]
          X.shape = [4, 1]
          X.data = [[1], [1], [3], [0]]
        set depth = 4
        Out is a LoDTensor:
          Out.lod = [[0, 1, 4]]
          Out.shape = [4, 4]
          Out.data = [[0., 1., 0., 0.],
                      [0., 1., 0., 0.],
                      [0., 0., 0., 1.],
                      [1., 0., 0., 0.]]
    """
    helper = LayerHelper("one_hot", **locals())
    one_hot_out = helper.create_tmp_variable(dtype='float32')
    helper.append_op(
        type="one_hot",
        inputs={'X': input},
        attrs={'depth': depth},
        outputs={'Out': one_hot_out})
    return one_hot_out
Y
Yu Yang 已提交
3977 3978


Y
Yu Yang 已提交
3979
def autoincreased_step_counter(counter_name=None, begin=1, step=1):
Y
Yu Yang 已提交
3980
    """
Y
Yu Yang 已提交
3981
    NOTE: The counter will be automatically increased by 1 every mini-batch
Y
Yu Yang 已提交
3982
    Return the run counter of the main program, which is started with 1.
Y
Yu Yang 已提交
3983 3984 3985 3986 3987 3988

    Args:
        counter_name(str): The counter name, default is '@STEP_COUNTER@'.
        begin(int): The first value of this counter.
        step(int): The increment step between each execution.

3989 3990
    Returns:
        Variable: The global run counter.
Y
Yu Yang 已提交
3991 3992
    """
    helper = LayerHelper('global_step_counter')
Y
Yu Yang 已提交
3993 3994
    if counter_name is None:
        counter_name = '@STEP_COUNTER@'
Y
Yu Yang 已提交
3995 3996 3997 3998 3999
    counter, is_new_var = helper.create_or_get_global_variable(
        name=counter_name, dtype='int64', shape=[1], persistable=True)
    if is_new_var:
        helper.set_variable_initializer(
            counter, initializer=Constant(
Y
Yu Yang 已提交
4000
                value=begin - 1, force_cpu=True))
Y
Yu Yang 已提交
4001 4002 4003
        helper.main_program.global_block().prepend_op(
            type='increment',
            inputs={'X': [counter]},
Y
Yu Yang 已提交
4004 4005
            outputs={'Out': [counter]},
            attrs={'step': float(step)})
Y
Yu Yang 已提交
4006 4007 4008
        counter.stop_gradient = True

    return counter
Y
yangyaming 已提交
4009 4010


4011
def reshape(x, shape, actual_shape=None, act=None, inplace=True, name=None):
C
caoying03 已提交
4012
    """
C
caoying03 已提交
4013 4014
    Gives a new shape to the input Tensor without changing its data.

4015 4016 4017 4018 4019
    The target shape can be given by :attr:`shape` or :attr:`actual_shape`.
    :attr:`shape` is a list of integer while :attr:`actual_shape` is a tensor
    variable. :attr:`actual_shape` has a higher priority than :attr:`shape`
    if it is provided, while :attr:`shape` still should be set correctly to
    gurantee shape inference in compile-time.
C
caoying03 已提交
4020

4021
    Some tricks exist when specifying the target shape.
C
caoying03 已提交
4022

4023 4024 4025 4026
    1. -1 means the value of this dimension is inferred from the total element
    number of x and remaining dimensions. Thus one and only one dimension can
    be set -1.

4027
    2. 0 means the actual dimension value is going to be copied from the
4028 4029 4030 4031
    corresponding dimension of x. The indice of 0s in shape can not exceed
    Rank(X).

    Here are some examples to explain it.
C
caoying03 已提交
4032 4033

    1. Given a 3-D tensor x with a shape [2, 4, 6], and the target shape
W
wanghaoshuang 已提交
4034
    is [6, 8], the reshape operator will transform x into a 2-D tensor with
4035
    shape [6, 8] and leaving x's data unchanged.
C
caoying03 已提交
4036

4037
    2. Given a 3-D tensor x with a shape [2, 4, 6], and the target shape
4038 4039
    specified is [2, 3, -1, 2], the reshape operator will transform x into a
    4-D tensor with shape [2, 3, 4, 2] and leaving x's data unchanged. In this
W
wanghaoshuang 已提交
4040 4041
    case, one dimension of the target shape is set to -1, the value of this
    dimension is inferred from the total element number of x and remaining
4042
    dimensions.
C
caoying03 已提交
4043

4044
    3. Given a 3-D tensor x with a shape [2, 4, 6], and the target shape
4045 4046 4047 4048
    is [-1, 0, 3, 2], the reshape operator will transform x into a 4-D tensor
    with shape [2, 4, 3, 2] and leaving x's data unchanged. In this case,
    besides -1, 0 means the actual dimension value is going to be copied from
    the corresponding dimension of x.
C
caoying03 已提交
4049 4050

    Args:
4051
        x(variable): The input tensor.
C
caoying03 已提交
4052 4053
        shape(list): The new shape. At most one dimension of the new shape can
                     be -1.
4054 4055 4056 4057 4058
        actual_shape(variable): An optional input. If provided, reshape
                                according to this given shape rather than
                                :attr:`shape` specifying shape. That is to
                                say :attr:`actual_shape` has a higher priority
                                than :attr:`shape`.
C
caoying03 已提交
4059 4060 4061 4062
        act (str): The non-linear activation to be applied to output variable.
        inplace(bool): If this flag is set true, a new output tensor is created
                       whose data is copied from input x, otherwise the output
                       shares data with input without copying.
4063
        name (str): The name of this layer. It is optional.
C
caoying03 已提交
4064

4065 4066
    Returns:
        Variable: The output tensor.
C
caoying03 已提交
4067 4068 4069

    Examples:
        .. code-block:: python
G
guosheng 已提交
4070

4071
            data = fluid.layers.data(
4072
                name='data', shape=[2, 4, 6], dtype='float32')
C
caoying03 已提交
4073
            reshaped = fluid.layers.reshape(
4074
                x=data, shape=[-1, 0, 3, 2], act='tanh', inplace=True)
C
caoying03 已提交
4075 4076 4077 4078 4079
    """

    if not (isinstance(shape, list) or isinstance(shape, tuple)):
        raise ValueError("Input shape must be a python lsit or tuple.")

4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094
    # Validate the shape
    unk_dim_idx = -1
    for dim_idx, dim_size in enumerate(shape):
        if dim_size == -1:
            assert unk_dim_idx == -1, (
                "Only one dimension in shape can be unknown.")
            unk_dim_idx = dim_idx
        elif dim_size == 0:
            assert dim_idx < len(x.shape), (
                "The indice of 0s in shape can not exceed Rank(X).")
        else:
            assert dim_size > 0, (
                "Each dimension size given in shape must not be negtive "
                "except one unknown dimension.")

C
caoying03 已提交
4095 4096 4097 4098
    helper = LayerHelper("reshape", **locals())
    reshaped = helper.create_tmp_variable(dtype=x.dtype)
    helper.append_op(
        type="reshape",
4099 4100 4101
        inputs={"X": x,
                "Shape": actual_shape}
        if isinstance(actual_shape, Variable) else {"X": x},
C
caoying03 已提交
4102 4103 4104 4105 4106
        attrs={"shape": shape,
               "inplace": inplace},
        outputs={"Out": reshaped})

    return helper.append_activation(reshaped)
4107 4108


Y
yangyaming 已提交
4109
def lod_reset(x, y=None, target_lod=None):
Y
yangyaming 已提交
4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201
    """
    LoD Reset Operator. Set LoD of **x** to a new one specified by **y** or
    **target_lod**. When **y** provided, **y.lod** would be considered as target
    LoD first, otherwise **y.data** would be considered as target LoD. If **y**
    is not provided, target LoD should be specified by **target_lod**.
    If target LoD is specified by **Y.data** or **target_lod**, only one level
    LoD is supported.

    .. code-block:: text

        * Example 1:

            Given a 1-level LoDTensor x:
                x.lod =  [[ 0,     2,                   5      6 ]]
                x.data = [[1.0], [2.0], [3.0], [4.0], [5.0], [6.0]]
                x.dims = [6, 1]

            target_lod: [0, 4, 6]

            then we get a 1-level LoDTensor:
                out.lod =  [[ 0,                   4,            6 ]]
                out.data = [[1.0], [2.0], [3.0], [4.0], [5.0], [6.0]]
                out.dims = [6, 1]

        * Example 2:

            Given a 1-level LoDTensor x:
                x.lod =  [[ 0,     2,                   5      6 ]]
                x.data = [[1.0], [2.0], [3.0], [4.0], [5.0], [6.0]]
                x.dims = [6, 1]

            y is a Tensor:
                y.data = [[0, 2, 6]]
                y.dims = [1, 3]

            then we get a 1-level LoDTensor:
                out.lod =  [[ 0,     2,                          6 ]]
                out.data = [[1.0], [2.0], [3.0], [4.0], [5.0], [6.0]]
                out.dims = [6, 1]

        * Example 3:

            Given a 1-level LoDTensor x:
                x.lod =  [[ 0,      2,                   5     6 ]]
                x.data = [[1.0], [2.0], [3.0], [4.0], [5.0], [6.0]]
                x.dims = [6, 1]

            y is a 2-level LoDTensor:
                y.lod =  [[0, 2, 4], [0, 2, 5, 6]]
                y.data = [[1.1], [2.1], [3.1], [4.1], [5.1], [6.1]]
                y.dims = [6, 1]

            then we get a 2-level LoDTensor:
                out.lod =  [[0, 2, 4], [0, 2, 5, 6]]
                out.data = [[1.0], [2.0], [3.0], [4.0], [5.0], [6.0]]
                out.dims = [6, 1]

    Args:
        x (Variable): Input variable which could be a Tensor or LodTensor.
        y (Variable|None): If provided, output's LoD would be derived from y.
        target_lod (list|tuple|None): One level LoD which should be considered
                                      as target LoD when y not provided.

    Returns:
        Variable: Output variable with LoD specified by this operator.

    Raises:
        ValueError: If y and target_lod are both None.

    Examples:
        .. code-block:: python

            x = layers.data(name='x', shape=[10])
            y = layers.data(name='y', shape=[10, 20], lod_level=2)
            out = layers.lod_reset(x=x, y=y)
    """
    helper = LayerHelper("lod_reset", **locals())
    out = helper.create_tmp_variable(dtype=x.dtype)
    if y is not None:
        helper.append_op(
            type="lod_reset", inputs={'X': x,
                                      'Y': y}, outputs={'Out': out})
    elif target_lod is not None:
        helper.append_op(
            type="lod_reset",
            inputs={'X': x},
            attrs={'target_lod': target_lod},
            outputs={'Out': out})
    else:
        raise ValueError("y and target_lod should not be both None.")

    return out
D
dragonwarrior 已提交
4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212


def lrn(input, n=5, k=1.0, alpha=1e-4, beta=0.75, name=None):
    """
    Local Response Normalization Layer. This layer performs a type of
    "lateral inhibition" by normalizing over local input regions.

    The formula is as follows:

    .. math::

D
dzhwinter 已提交
4213
      Output(i, x, y) = Input(i, x, y) / \left( \\
D
dzhwinter 已提交
4214
        k + \alpha \sum\limits^{\min(C, c + n/2)}_{j = \max(0, c - n/2)} \\
D
dzhwinter 已提交
4215
        (Input(j, x, y))^2\right)^{\beta}
D
dragonwarrior 已提交
4216 4217 4218 4219 4220 4221 4222 4223 4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243

    In the above equation:

    * :math:`n`: The number of channels to sum over.
    * :math:`k`: The offset (avoid being divided by 0).
    * :math:`alpha`: The scaling parameter.
    * :math:`beta`: The exponent parameter.

    Refer to `ImageNet Classification with Deep Convolutional Neural Networks
    <https://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks.pdf>`_

    Args:
        input (Variable): The input tensor of this layer, and the dimension of input tensor must be 4.
        n (int, default 5): The number of channels to sum over.
        k (float, default 1.0): An offset (usually positive to avoid dividing by 0).
        alpha (float, default 1e-4): The scaling parameter.
        beta (float, default 0.75): The exponent.
        name (str, default None): A name for this operation.

    Raises:
        ValueError: If rank of the input tensor is not 4.

    Returns:
        A tensor variable storing the transformation result.

    Examples:
        .. code-block:: python

F
stash  
fengjiayi 已提交
4244 4245
          data = fluid.layers.data(
              name="data", shape=[3, 112, 112], dtype="float32")
D
dragonwarrior 已提交
4246 4247 4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272
          lrn = fluid.layers.lrn(input=data)
    """
    helper = LayerHelper('lrn', **locals())
    dtype = helper.input_dtype()
    input_shape = input.shape
    dims = len(input_shape)

    if dims != 4:
        raise ValueError(
            "dims of input must be 4(not %d), and it's order must be NCHW" %
            (dims))

    mid_out = helper.create_tmp_variable(dtype=dtype, stop_gradient=True)
    lrn_out = helper.create_tmp_variable(dtype)
    helper.append_op(
        type="lrn",
        inputs={"X": input},
        outputs={
            "Out": lrn_out,
            "MidOut": mid_out,
        },
        attrs={"n": n,
               "k": k,
               "alpha": alpha,
               "beta": beta})

    return lrn_out
G
guosheng 已提交
4273 4274 4275 4276


def pad(x, paddings, pad_value=0., name=None):
    """
G
guosheng 已提交
4277
    Pads a tensor with a constant value given by :attr:`pad_value`, and the
W
wanghaoshuang 已提交
4278
    padded width is specified by :attr:`paddings`.
G
guosheng 已提交
4279

G
guosheng 已提交
4280 4281 4282 4283
    Specifically, the number of values padded before the contents of :attr:`x`
    in dimension :attr:`i` is indicated by :attr:`paddings[i]`, and the number
    of values padded after the contents of :attr:`x` in dimension :attr:`i` is
    indicated by :attr:`paddings[i+1]`.
G
guosheng 已提交
4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305

    See below for an example.

    .. code-block:: text

        Given:
            x = [[1, 2], [3, 4]]

            paddings = [0, 1, 1, 2]

            pad_value = 0

        Return:

            out = [[0, 1, 2, 0, 0]
                   [0, 3, 4, 0, 0]
                   [0, 0, 0, 0, 0]]

    Args:
        x (Variable): The input tensor variable.
        paddings (list): A list of integers. Its elements specify the padded
                         width before and after for each dimension in turn.
W
wanghaoshuang 已提交
4306
                         The length of :attr:paddings must be
G
guosheng 已提交
4307 4308 4309 4310 4311 4312 4313 4314 4315 4316
                         :math:`rank(x) \\times 2`.
        pad_value (float): The constant value used to pad.
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.

    Returns:
        Variable: The padded tensor variable.

    Examples:
        .. code-block:: python
G
guosheng 已提交
4317

G
guosheng 已提交
4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331
            # x is a rank 2 tensor variable.
            out = fluid.layers.pad(
                x=x, paddings=[0, 1, 1, 2], pad_value=0.)
    """
    helper = LayerHelper('pad', input=x, **locals())
    dtype = helper.input_dtype()
    out = helper.create_tmp_variable(dtype)
    helper.append_op(
        type='pad',
        inputs={'X': x},
        outputs={'Out': out},
        attrs={'paddings': paddings,
               'pad_value': float(pad_value)})
    return out
4332 4333 4334 4335 4336 4337 4338 4339 4340


def label_smooth(label,
                 prior_dist=None,
                 epsilon=0.1,
                 dtype="float32",
                 name=None):
    """
    Label smoothing is a mechanism to regularize the classifier layer and is
4341 4342
    called label-smoothing regularization (LSR).

4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365
    Label smoothing is proposed to encourage the model to be less confident,
    since optimizing the log-likelihood of the correct label directly may
    cause overfitting and reduce the ability of the model to adapt. Label
    smoothing replaces the ground-truth label :math:`y` with the weighted sum
    of itself and some fixed distribution :math:`\mu`. For class :math:`k`,
    i.e.

    .. math::

        \\tilde{y_k} = (1 - \epsilon) * y_k + \epsilon * \mu_k,

    where :math:`1 - \epsilon` and :math:`\epsilon` are the weights
    respectively, and :math:`\\tilde{y}_k` is the smoothed label. Usually
    uniform distribution is used for :math:`\mu`.

    See more details about label smoothing in https://arxiv.org/abs/1512.00567.

    Args:
        label(Variable): The input variable containing the label data. The
                          label data should use one-hot representation.
        prior_dist(Variable): The prior distribution to be used to smooth
                              labels. If not provided, an uniform distribution
                              is used. The shape of :attr:`prior_dist` should
4366
                              be :math:`(1, class\_num)`.
4367 4368
        epsilon(float): The weight used to mix up the original ground-truth
                        distribution and the fixed distribution.
4369
        dtype(np.dtype|core.VarDesc.VarType|str): The type of data : float32,
4370 4371 4372 4373 4374 4375 4376 4377 4378 4379 4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396
                                                  float_64, int etc.
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.

    Returns:
        Variable: The tensor variable containing the smoothed labels.

    Examples:
        .. code-block:: python

            label = layers.data(name="label", shape=[1], dtype="float32")
            one_hot_label = layers.one_hot(input=label, depth=10)
            smooth_label = layers.label_smooth(
                label=one_hot_label, epsilon=0.1, dtype="float32")
    """
    if epsilon > 1. or epsilon < 0.:
        raise ValueError("The value of epsilon must be between 0 and 1.")
    helper = LayerHelper("label_smooth", **locals())
    label.stop_gradient = True
    smooth_label = helper.create_tmp_variable(dtype)
    helper.append_op(
        type="label_smooth",
        inputs={"X": label,
                "PriorDist": prior_dist} if prior_dist else {"X": label},
        outputs={"Out": smooth_label},
        attrs={"epsilon": float(epsilon)})
    return smooth_label
4397 4398 4399 4400


def roi_pool(input, rois, pooled_height=1, pooled_width=1, spatial_scale=1.0):
    """
4401
    Region of interest pooling (also known as RoI pooling) is to perform
4402 4403
        is to perform max pooling on inputs of nonuniform sizes to obtain
        fixed-size feature maps (e.g. 7*7).
4404 4405 4406 4407
    The operator has three steps:
        1. Dividing each region proposal into equal-sized sections with
           the pooled_width and pooled_height
        2. Finding the largest value in each section
4408 4409 4410 4411 4412 4413 4414
        3. Copying these max values to the output buffer

    Args:
        input (Variable): The input for ROI pooling.
        rois (Variable): ROIs (Regions of Interest) to pool over. It should
                         be a 2-D one level LoTensor of shape [num_rois, 4].
                         The layout is [x1, y1, x2, y2], where (x1, y1)
4415 4416
                         is the top left coordinates, and (x2, y2) is the
                         bottom right coordinates. The num_rois is the
4417 4418 4419 4420 4421 4422 4423 4424
                         total number of ROIs in this batch data.
        pooled_height (integer): The pooled output height. Default: 1
        pooled_width (integer): The pooled output width. Default: 1
        spatial_scale (float): Multiplicative spatial scale factor. To
                               translate ROI coords from their input scale
                               to the scale used when pooling. Default: 1.0

    Returns:
4425
        pool_out (Variable): The output is a 4-D tensor of the shape
4426 4427 4428
                             (num_rois, channels, pooled_h, pooled_w).

    Examples:
4429 4430
        .. code-block:: python

4431
            pool_out = fluid.layers.roi_pool(input=x, rois=rois, 7, 7, 1.0)
4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448
    """
    helper = LayerHelper('roi_pool', **locals())
    dtype = helper.input_dtype()
    pool_out = helper.create_tmp_variable(dtype)
    argmaxes = helper.create_tmp_variable(dtype='int32')
    helper.append_op(
        type="roi_pool",
        inputs={"X": input,
                "ROIs": rois},
        outputs={"Out": pool_out,
                 "Argmax": argmaxes},
        attrs={
            "pooled_height": pooled_height,
            "pooled_width": pooled_width,
            "spatial_scale": spatial_scale
        })
    return pool_out
W
whs 已提交
4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463 4464 4465 4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476


def dice_loss(input, label, epsilon=0.00001):
    """
    Dice loss for comparing the similarity of two batch of data,
    usually is used for binary image segmentation i.e. labels are binary.
    The dice loss can be defined as below equation:

    .. math::

        dice\_loss &= 1 - \\frac{2 * intersection\_area}{total\_area} \\\\
                  &= \\frac{(total\_area - intersection\_area) - intersection\_area}{total\_area} \\\\
                  &= \\frac{(union\_area - intersection\_area)}{total\_area}


    Args:
        input (Variable): The predictions with rank>=2. The first dimension is batch size,
                          and the last dimension is class number.
        label (Variable): The groud truth with the same rank with input. The first dimension
                          is batch size, and the last dimension is 1.
        epsilon (float): The epsilon will be added to the numerator and denominator.
                         If both input and label are empty, it makes sure dice is 1.
                         Default: 0.00001

    Returns:
        dice_loss (Variable): The dice loss with shape [1].

    Examples:
4477 4478
        .. code-block:: python

W
whs 已提交
4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489
            predictions = fluid.layers.softmax(x)
            loss = fluid.layers.dice_loss(input=predictions, label=label, 2)
    """
    label = one_hot(label, depth=input.shape[-1])
    reduce_dim = range(1, len(input.shape))
    inse = reduce_sum(input * label, dim=reduce_dim)
    dice_denominator = reduce_sum(
        input, dim=reduce_dim) + reduce_sum(
            label, dim=reduce_dim)
    dice_score = 1 - inse * 2 / (dice_denominator + epsilon)
    return reduce_mean(dice_score)
4490 4491


4492 4493 4494 4495 4496
def image_resize(input,
                 out_shape=None,
                 scale=None,
                 name=None,
                 resample='BILINEAR'):
4497
    """
4498
    Resize a batch of images.
F
stash  
fengjiayi 已提交
4499

4500 4501 4502 4503 4504
    The input must be a tensor of the shape (num_batches, channels, in_h, in_w), 
    and the resizing only applies on the last two dimensions(hight and width).

    Supporting resample methods:
        'BILINEAR' : Bilinear interpolation
F
stash  
fengjiayi 已提交
4505

4506
    Args:
4507
        input (Variable): The input tensor of image resize layer,
4508 4509
                          This is a 4-D tensor of the shape
                          (num_batches, channels, in_h, in_w).
4510
        out_shape(list|tuple|Variable|None): Output shape of image resize
4511 4512
                                    layer, the shape is (out_h, out_w).
                                    Default: None
B
baiyf 已提交
4513
        scale(float|None): The multiplier for the input height or width.
4514 4515 4516
                         At least one of out_shape or scale must be set.
                         And out_shape has a higher priority than scale.
                         Default: None
4517 4518
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.
4519 4520
        resample(str): The resample method. It can only be 'BILINEAR' currently.
                       Default: 'BILINEAR'
4521 4522 4523 4524

    Returns:
        out (Variable): The output is a 4-D tensor of the shape
                        (num_batches, channls, out_h, out_w).
F
stash  
fengjiayi 已提交
4525

4526 4527 4528
    Examples:
        .. code-block:: python

4529
            out = fluid.layers.image_resize(input, out_shape=[12, 12])
4530
    """
4531 4532 4533 4534
    resample_methods = {'BILINEAR': 'bilinear_interp'}
    if resample not in resample_methods:
        raise ValueError(
            "The 'resample' of image_resize can only be 'BILINEAR' currently.")
4535 4536
    if out_shape is None and scale is None:
        raise ValueError("One of out_shape and scale must not be None")
4537 4538
    helper = LayerHelper('bilinear_interp', **locals())
    dtype = helper.input_dtype()
4539 4540 4541 4542

    def _is_list_or_turple_(data):
        return (isinstance(data, list) or isinstance(data, tuple))

4543 4544 4545
    out_h = 0
    out_w = 0
    inputs = {"X": input}
4546
    if out_shape is not None:
B
baiyf 已提交
4547 4548 4549
        if not (_is_list_or_turple_(out_shape) and
                len(out_shape) == 2) and not isinstance(out_shape, Variable):
            raise ValueError('out_shape should be a list or tuple or variable')
4550 4551 4552 4553 4554 4555
        if _is_list_or_turple_(out_shape):
            out_shape = list(map(int, out_shape))
            out_h = out_shape[0]
            out_w = out_shape[1]
        else:
            inputs['OutSize'] = out_shape
4556 4557 4558 4559
    else:
        out_h = int(input.shape[2] * scale)
        out_w = int(input.shape[3] * scale)

4560 4561
    out = helper.create_tmp_variable(dtype)
    helper.append_op(
4562
        type=resample_methods[resample],
4563
        inputs=inputs,
4564 4565 4566 4567
        outputs={"Out": out},
        attrs={"out_h": out_h,
               "out_w": out_w})
    return out
F
stash  
fengjiayi 已提交
4568 4569


Y
yuyang18 已提交
4570
@templatedoc(op_type="bilinear_interp")
4571 4572
def resize_bilinear(input, out_shape=None, scale=None, name=None):
    """
Y
yuyang18 已提交
4573 4574 4575 4576 4577 4578
    ${comment}

    Args:
        input(${x_type}): ${x_comment}.

        out_shape(${out_size_type}): ${out_size_comment}.
4579

Y
yuyang18 已提交
4580 4581 4582 4583 4584 4585 4586 4587
        scale(float|None): The multiplier for the input height or width. At
             least one of out_shape or scale must be set. And out_shape has
             a higher priority than scale. Default: None.

        name(str|None): The output variable name.

    Returns:
        ${out_comment}.
4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604
    """

    return image_resize(input, out_shape, scale, name, 'BILINEAR')


def image_resize_short(input, out_short_len, resample='BILINEAR'):
    """
    Resize a batch of images. The short edge of input images will be 
    resized to the given 'out_short_len'. The long edge of input images 
    will be resized proportionately to make images' length-width ratio 
    constant.

    Args:
        input (Variable): The input tensor of image resize layer,
                          This is a 4-D tensor of the shape
                          (num_batches, channels, in_h, in_w).
        out_short_len(int): The length of output images' short edge.
4605
        resample (str): resample method, default: BILINEAR.
F
fengjiayi 已提交
4606

4607 4608 4609 4610 4611 4612 4613 4614 4615 4616 4617 4618 4619
    Returns:
        out (Variable): The output is a 4-D tensor of the shape
                        (num_batches, channls, out_h, out_w).
    """
    in_shape = input.shape
    if len(in_shape) != 4:
        raise ValueError(
            "The rank of input must be 4 (num_batches, channels, in_h, in_w).")
    hw = in_shape[2:4]
    short_idx = hw.index(min(hw))
    long_idx = 1 - short_idx
    out_shape = list(hw)
    out_shape[short_idx] = out_short_len
F
fengjiayi 已提交
4620 4621 4622
    out_shape[long_idx] = int(
        float(out_shape[long_idx]) * (float(out_short_len) / float(hw[
            short_idx])) + 0.5)
4623 4624 4625
    return image_resize(input=input, out_shape=out_shape, resample=resample)


W
whs 已提交
4626 4627 4628 4629 4630 4631 4632
def gather(input, index):
    """
    Output is obtained by gathering entries of the outer-most dimension 
    of X indexed by `index` and concatenate them together.

    .. math::

4633
        Out = X[Index]
W
whs 已提交
4634 4635 4636 4637 4638 4639 4640


    .. code-block:: text


                Given:

4641 4642
                X = [[1, 2],
                     [3, 4],
W
whs 已提交
4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659
                     [5, 6]]

                Index = [1, 2]

                Then:

                Out = [[3, 4],
                       [5, 6]]

    Args:
        input (Variable): The source input with rank>=1. 
        index (Variable): The index input with rank=1.

    Returns:
        output (Variable): The output is a tensor with the same rank as input.

    Examples:
W
whs 已提交
4660

W
whs 已提交
4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675
        .. code-block:: python

            output = fluid.layers.gather(x, index)
    """
    helper = LayerHelper('gather', **locals())
    dtype = helper.input_dtype()
    out = helper.create_tmp_variable(dtype)
    helper.append_op(
        type="gather",
        inputs={"X": input,
                "Index": index},
        outputs={"Out": out})
    return out


Y
yuyang18 已提交
4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694
@templatedoc()
def random_crop(x, shape, seed=None):
    """
    ${comment}

    Examples:
        >>> img = fluid.layers.data("img", [3, 256, 256])
        >>> cropped_img = fluid.layers.random_crop(img, shape=[3, 224, 224])

    Args:
        x(${x_type}): ${x_comment}
        shape(${shape_type}): ${shape_comment}
        seed(int|${seed_type}|None): ${seed_comment} By default, the seed will
            get from `random.randint(-65536, 65535)`.

    Returns:
        ${out_comment}

    """
F
stash  
fengjiayi 已提交
4695 4696 4697
    helper = LayerHelper("random_crop", **locals())
    dtype = helper.input_dtype()
    out = helper.create_tmp_variable(dtype)
Y
yuyang18 已提交
4698 4699 4700
    if seed is None:
        seed = random.randint(-65536, 65535)

F
stash  
fengjiayi 已提交
4701
    if isinstance(seed, int):
F
fengjiayi 已提交
4702
        seed_value = seed
F
fengjiayi 已提交
4703 4704 4705 4706 4707 4708 4709 4710
        seed = helper.create_tmp_variable(dtype="int64")
        helper.append_op(
            type="fill_constant",
            inputs={},
            outputs={"Out": seed},
            attrs={
                "dtype": seed.dtype,
                "shape": [1],
F
fengjiayi 已提交
4711 4712
                "value": float(seed_value),
                "force_cpu": True
F
fengjiayi 已提交
4713
            })
F
stash  
fengjiayi 已提交
4714 4715
    elif not isinstance(seed, Variable):
        raise ValueError("'seed' must be a Variable or an int.")
F
fengjiayi 已提交
4716
    seed_out = helper.create_tmp_variable(dtype="int64")
F
stash  
fengjiayi 已提交
4717 4718
    helper.append_op(
        type="random_crop",
F
fix  
fengjiayi 已提交
4719
        inputs={"X": x,
F
stash  
fengjiayi 已提交
4720 4721 4722 4723 4724
                "Seed": seed},
        outputs={"Out": out,
                 "SeedOut": seed_out},
        attrs={"shape": shape})
    return out
W
whs 已提交
4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774


def mean_iou(input, label, num_classes):
    """
    Mean Intersection-Over-Union is a common evaluation metric for
    semantic image segmentation, which first computes the IOU for each 
    semantic class and then computes the average over classes. 
    IOU is defined as follows: 
    
    .. math::
        
        IOU = true_positive / (true_positive + false_positive + false_negative). 

    The predictions are accumulated in a confusion matrix and mean-IOU 
    is then calculated from it.


    Args:
        input (Variable): A Tensor of prediction results for semantic labels with type int32 or int64.
        label (Variable):  A Tensor of ground truth labels with type int32 or int64. 
                           Its shape should be the same as input.

    Returns:
        mean_iou (Variable): A Tensor representing the mean intersection-over-union with shape [1].
        out_wrong(Variable): A Tensor with shape [num_classes]. The wrong numbers of each class.
        out_correct(Variable): A Tensor with shape [num_classes]. The correct numbers of each class. 


    Examples:

        .. code-block:: python

            iou, wrongs, corrects = fluid.layers.mean_iou(predict, label, num_classes)
    """
    helper = LayerHelper('mean_iou', **locals())
    dtype = helper.input_dtype()
    out_mean_iou = helper.create_tmp_variable(dtype='float32')
    out_wrong = helper.create_tmp_variable(dtype='int32')
    out_correct = helper.create_tmp_variable(dtype='int32')
    helper.append_op(
        type="mean_iou",
        inputs={"predictions": input,
                "labels": label},
        outputs={
            "out_mean_iou": out_mean_iou,
            "out_wrong": out_wrong,
            "out_correct": out_correct
        },
        attrs={"num_classes": num_classes})
    return out_mean_iou, out_wrong, out_correct