nn.py 148.5 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 16 17 18 19 20
"""
All layers just related to the neural network.
"""

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
22
from layer_function_generator import autodoc
Y
yangyaming 已提交
23
from tensor import concat
C
chengduoZH 已提交
24
import utils
Y
Yu Yang 已提交
25 26

__all__ = [
Y
ying 已提交
27 28 29
    'fc',
    'embedding',
    'dynamic_lstm',
Y
Yibing Liu 已提交
30
    'dynamic_lstmp',
G
guosheng 已提交
31
    'dynamic_gru',
Y
ying 已提交
32 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',
    'sequence_pool',
42 43
    'sequence_softmax',
    'softmax',
Y
ying 已提交
44 45 46 47 48 49 50 51 52 53
    'pool2d',
    'batch_norm',
    'beam_search_decode',
    'conv2d_transpose',
    'sequence_expand',
    'lstm_unit',
    'reduce_sum',
    'reduce_mean',
    'reduce_max',
    'reduce_min',
54
    'reduce_prod',
Y
ying 已提交
55 56 57 58
    'sequence_first_step',
    'sequence_last_step',
    'dropout',
    'split',
59 60
    'ctc_greedy_decoder',
    'edit_distance',
Y
ying 已提交
61 62
    'l2_normalize',
    'matmul',
Q
qingqing01 已提交
63
    'topk',
Y
ying 已提交
64 65
    'warpctc',
    'sequence_reshape',
66
    'transpose',
67
    'im2sequence',
68
    'nce',
Q
Qiao Longfei 已提交
69
    'beam_search',
70
    'row_conv',
71
    'multiplex',
G
guosheng 已提交
72
    'layer_norm',
73 74
    'softmax_with_cross_entropy',
    'smooth_l1',
75
    'one_hot',
Y
Yu Yang 已提交
76
    'autoincreased_step_counter',
C
caoying03 已提交
77
    'reshape',
Y
yangyaming 已提交
78
    'lod_reset',
D
dragonwarrior 已提交
79
    'lrn',
G
guosheng 已提交
80
    'pad',
81
    'label_smooth',
82
    'roi_pool',
W
whs 已提交
83
    'dice_loss',
84
    'upsampling_bilinear2d',
85
    'random_crop',
Y
Yu Yang 已提交
86 87 88 89 90 91 92 93
]


def fc(input,
       size,
       num_flatten_dims=1,
       param_attr=None,
       bias_attr=None,
K
Kexin Zhao 已提交
94
       use_cudnn=False,
95
       use_mkldnn=False,
Y
Yu Yang 已提交
96
       act=None,
J
Jacek Czaja 已提交
97
       is_test=False,
98
       name=None):
Y
Yu Yang 已提交
99
    """
100
    **Fully Connected Layer**
Y
Yu Yang 已提交
101

C
caoying03 已提交
102
    The fully connected layer can take multiple tensors as its inputs. It
R
ranqiu 已提交
103 104 105 106 107 108
    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
109
    if activation is not None, it will be applied to the output as well.
C
caoying03 已提交
110

C
caoying03 已提交
111
    This process can be formulated as follows:
112 113 114

    .. math::

115
        Out = Act({\sum_{i=0}^{N-1}X_iW_i + b})
116 117 118

    In the above equation:

C
caoying03 已提交
119 120 121 122
    * :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).
123
    * :math:`Act`: The activation function.
C
caoying03 已提交
124
    * :math:`Out`: The output tensor.
Y
Yu Yang 已提交
125 126

    Args:
R
ranqiu 已提交
127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143
        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
144
        is_test(bool): A flag indicating whether execution is in test phase.
M
mozga-intel 已提交
145 146
        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
147
        name (str, default None): The name of this layer.
Y
Yu Yang 已提交
148

149
    Returns:
R
ranqiu 已提交
150
        A tensor variable storing the transformation result.
151 152

    Raises:
C
caoying03 已提交
153
        ValueError: If rank of the input tensor is less than 2.
154 155 156 157

    Examples:
        .. code-block:: python

F
stash  
fengjiayi 已提交
158 159
          data = fluid.layers.data(
              name="data", shape=[32, 32], dtype="float32")
160
          fc = fluid.layers.fc(input=data, size=1000, act="tanh")
Y
Yu Yang 已提交
161
    """
C
caoying03 已提交
162

C
caoying03 已提交
163
    helper = LayerHelper("fc", **locals())
Y
Yu Yang 已提交
164 165 166 167

    dtype = helper.input_dtype()

    mul_results = []
168 169
    for input_var, param_attr in helper.iter_inputs_and_params():
        input_shape = input_var.shape
Y
Yu Yang 已提交
170 171 172
        param_shape = [
            reduce(lambda a, b: a * b, input_shape[num_flatten_dims:], 1)
        ] + [size]
Y
ying 已提交
173

Y
Yu Yang 已提交
174
        w = helper.create_parameter(
175 176
            attr=param_attr, shape=param_shape, dtype=dtype, is_bias=False)
        tmp = helper.create_tmp_variable(dtype)
177
        helper.append_op(
178 179 180
            type="mul",
            inputs={"X": input_var,
                    "Y": w},
181
            outputs={"Out": tmp},
J
Jacek Czaja 已提交
182
            attrs={
183 184 185
                "x_num_col_dims": num_flatten_dims,
                "y_num_col_dims": 1,
                "use_mkldnn": use_mkldnn
J
Jacek Czaja 已提交
186
            })
187 188 189 190
        mul_results.append(tmp)

    if len(mul_results) == 1:
        pre_bias = mul_results[0]
191
    else:
192 193 194 195 196 197 198
        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
199 200


201 202 203
def embedding(input,
              size,
              is_sparse=False,
204
              is_distributed=False,
205 206 207
              padding_idx=None,
              param_attr=None,
              dtype='float32'):
Y
Yu Yang 已提交
208
    """
209 210
    **Embedding Layer**

211
    This layer is used to lookup embeddings of IDs, provided by :attr:`input`, in
212 213
    a lookup table. The result of this lookup is the embedding of each ID in the
    :attr:`input`.
214 215 216

    All the input variables are passed in as local variables to the LayerHelper
    constructor.
Y
Yu Yang 已提交
217 218

    Args:
219 220 221 222 223 224 225
        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.
        padding_idx(int|long|None): If :attr:`None`, it makes no effect to lookup.
            Otherwise the given :attr:`padding_idx` indicates padding the output
226 227
            with zeros whenever lookup encounters it in :attr:`input`. If
            :math:`padding_idx < 0`, the padding_idx to use in lookup is
228 229
            :math:`size[0] + dim`.
        param_attr(ParamAttr): Parameters for this layer
230
        dtype(np.dtype|core.VarDesc.VarType|str): The type of data : float32, float_16, int etc
Y
Yu Yang 已提交
231

232 233 234
    Returns:
        Variable: The tensor variable storing the embeddings of the \
                  supplied inputs.
Y
Yu Yang 已提交
235

236 237
    Examples:
        .. code-block:: python
Y
Yu Yang 已提交
238

C
chengduoZH 已提交
239
          dict_size = len(dataset.ids)
240
          data = fluid.layers.data(name='ids', shape=[32, 32], dtype='float32')
C
chengduoZH 已提交
241
          fc = fluid.layers.embedding(input=data, size=[dict_size, 16])
Y
Yu Yang 已提交
242 243 244 245 246 247
    """

    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)
248 249
    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
250 251 252 253 254
    helper.append_op(
        type='lookup_table',
        inputs={'Ids': input,
                'W': w},
        outputs={'Out': tmp},
255 256 257 258 259
        attrs={
            'is_sparse': is_sparse,
            'is_distributed': is_distributed,
            'padding_idx': padding_idx
        })
Y
Yu Yang 已提交
260 261 262 263 264 265 266 267 268 269 270 271 272
    return tmp


# TODO(qijun): expose H0 and C0
def dynamic_lstm(input,
                 size,
                 param_attr=None,
                 bias_attr=None,
                 use_peepholes=True,
                 is_reverse=False,
                 gate_activation='sigmoid',
                 cell_activation='tanh',
                 candidate_activation='tanh',
273 274
                 dtype='float32',
                 name=None):
Y
Yibing Liu 已提交
275 276 277 278 279 280
    """
    **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
281
    .. math::
Y
Yibing Liu 已提交
282

283
        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
284

285
        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
286

287
        \\tilde{c_t} & = act_g(W_{cx}x_t + W_{ch}h_{t-1} + b_c)
Y
Yibing Liu 已提交
288

289 290 291
        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
292

Y
Yibing Liu 已提交
293
        h_t & = o_t \odot act_h(c_t)
Y
Yibing Liu 已提交
294

295
    where the :math:`W` terms denote weight matrices (e.g. :math:`W_{xi}` is
296
    the matrix of weights from the input gate to the input), :math:`W_{ic}, \
297 298 299
    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
300
    gate bias vector), :math:`\sigma` is the non-linear activations, such as
301 302
    logistic sigmoid function, and :math:`i, f, o` and :math:`c` are the input
    gate, forget gate, output gate, and cell activation vectors, respectively,
303 304
    all of which have the same size as the cell output activation vector :math:`h`.

305 306 307 308
    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
309 310 311
    the previous hidden state.

    Set `use_peepholes` to `False` to disable peephole connection. The formula
Y
Yibing Liu 已提交
312 313 314
    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
315 316 317
    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
318 319

    Args:
320 321 322 323
        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
324 325
                         mini-batch, D is the hidden size.
        size(int): 4 * hidden size.
326
        param_attr(ParamAttr|None): The parameter attribute for the learnable
327
                               hidden-hidden weights.
Y
Yibing Liu 已提交
328 329 330

                               - Weights = {:math:`W_{ch}, W_{ih}, \
                                                W_{fh}, W_{oh}`}
331 332 333
                               - The shape is (D x 4D), where D is the hidden
                                 size.
        bias_attr(ParamAttr|None): The bias attribute for the learnable bias
334 335 336
                              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
337

338
                              1. `use_peepholes = False`
Y
Yibing Liu 已提交
339
                                - Biases = {:math:`b_c, b_i, b_f, b_o`}.
340
                                - The shape is (1 x 4D).
341
                              2. `use_peepholes = True`
Y
Yibing Liu 已提交
342 343
                                - Biases = { :math:`b_c, b_i, b_f, b_o, W_{ic}, \
                                                 W_{fc}, W_{oc}`}.
344
                                - The shape is (1 x 7D).
345
        use_peepholes(bool): Whether to enable diagonal/peephole connections,
Y
Yibing Liu 已提交
346 347
                             default `True`.
        is_reverse(bool): Whether to compute reversed LSTM, default `False`.
348 349
        gate_activation(str): The activation for input gate, forget gate and
                              output gate. Choices = ["sigmoid", "tanh", "relu",
Y
Yibing Liu 已提交
350
                              "identity"], default "sigmoid".
351
        cell_activation(str): The activation for cell output. Choices = ["sigmoid",
Y
Yibing Liu 已提交
352 353
                              "tanh", "relu", "identity"], default "tanh".
        candidate_activation(str): The activation for candidate hidden state.
F
stash  
fengjiayi 已提交
354 355
                              Choices = ["sigmoid", "tanh",
                                  "relu", "identity"],
Y
Yibing Liu 已提交
356 357
                              default "tanh".
        dtype(str): Data type. Choices = ["float32", "float64"], default "float32".
358 359
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.
Y
Yibing Liu 已提交
360 361

    Returns:
Y
Yibing Liu 已提交
362 363
        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
364

Y
Yibing Liu 已提交
365
    Examples:
Y
Yibing Liu 已提交
366 367
        .. code-block:: python

Y
Yibing Liu 已提交
368 369
            hidden_dim = 512
            forward_proj = fluid.layers.fc(input=input_seq, size=hidden_dim * 4,
370
                                           act=None, bias_attr=None)
Y
Yibing Liu 已提交
371 372
            forward, _ = fluid.layers.dynamic_lstm(
                input=forward_proj, size=hidden_dim * 4, use_peepholes=False)
Y
Yibing Liu 已提交
373
    """
374

Y
Yu Yang 已提交
375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410
    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)

    helper.append_op(
        type='lstm',
        inputs={'Input': input,
                'Weight': weight,
                'Bias': bias},
        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
411 412 413 414 415 416 417 418 419 420 421
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',
422 423
                  dtype='float32',
                  name=None):
Y
Yibing Liu 已提交
424 425 426
    """
    **Dynamic LSTMP Layer**

427 428 429 430 431 432
    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
433 434 435 436 437

    The formula is as follows:

    .. math::

438
        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
439

440
        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
441

442
        \\tilde{c_t} & = act_g(W_{cx}x_t + W_{cr}r_{t-1} + b_c)
Y
Yibing Liu 已提交
443

444
        o_t & = \sigma(W_{ox}x_{t} + W_{or}r_{t-1} + W_{oc}c_t + b_o)
Y
Yibing Liu 已提交
445

446
        c_t & = f_t \odot c_{t-1} + i_t \odot \\tilde{c_t}
Y
Yibing Liu 已提交
447

448
        h_t & = o_t \odot act_h(c_t)
Y
Yibing Liu 已提交
449

450
        r_t & = \overline{act_h}(W_{rh}h_t)
Y
Yibing Liu 已提交
451

Y
Yibing Liu 已提交
452 453 454 455 456 457
    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, \
458
          we use vectors to reprenset these diagonal weight matrices.
Y
Yibing Liu 已提交
459
    * :math:`b`: Denotes bias vectors (e.g. :math:`b_i` is the input gate \
460
          bias vector).
Y
Yibing Liu 已提交
461 462 463
    * :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 \
464
          the same size as the cell output activation vector :math:`h`.
Y
Yibing Liu 已提交
465
    * :math:`h`: The hidden state.
466
    * :math:`r`: The recurrent projection of the hidden state.
Y
Yibing Liu 已提交
467 468
    * :math:`\\tilde{c_t}`: The candidate hidden state, whose \
          computation is based on the current input and previous hidden state.
469
    * :math:`\odot`: The element-wise product of the vectors.
Y
Yibing Liu 已提交
470
    * :math:`act_g` and :math:`act_h`: The cell input and cell output \
471
          activation functions and `tanh` is usually used for them.
Y
Yibing Liu 已提交
472 473
    * :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
474 475 476 477

    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.
478

Y
Yibing Liu 已提交
479 480 481 482 483 484 485 486 487 488 489 490
    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.
491
        param_attr(ParamAttr|None): The parameter attribute for the learnable
Y
Yibing Liu 已提交
492 493
                               hidden-hidden weight and projection weight.

494 495
                               - Hidden-hidden weight = {:math:`W_{ch}, W_{ih}, \
                                                W_{fh}, W_{oh}`}.
496 497
                               - 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
498 499
                                 size.
                               - Projection weight = {:math:`W_{rh}`}.
500 501
                               - 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
502 503 504 505 506 507
                              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`}.
508
                                - The shape is (1 x 4D).
Y
Yibing Liu 已提交
509 510 511
                              2. `use_peepholes = True`
                                - Biases = { :math:`b_c, b_i, b_f, b_o, W_{ic}, \
                                                 W_{fc}, W_{oc}`}.
512
                                - The shape is (1 x 7D).
Y
Yibing Liu 已提交
513 514 515 516 517 518 519 520 521
        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
522 523
                              Choices = ["sigmoid", "tanh",
                                  "relu", "identity"],
Y
Yibing Liu 已提交
524 525
                              default "tanh".
        proj_activation(str): The activation for projection output.
F
stash  
fengjiayi 已提交
526 527
                              Choices = ["sigmoid", "tanh",
                                  "relu", "identity"],
Y
Yibing Liu 已提交
528 529
                              default "tanh".
        dtype(str): Data type. Choices = ["float32", "float64"], default "float32".
530 531
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.
Y
Yibing Liu 已提交
532 533

    Returns:
534 535
        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
536 537 538 539 540
               (T x D), and both LoD is the same with the `input`.

    Examples:
        .. code-block:: python

Y
Yibing Liu 已提交
541
            hidden_dim, proj_dim = 512, 256
Y
Yibing Liu 已提交
542 543
            fc_out = fluid.layers.fc(input=input_seq, size=hidden_dim * 4,
                                     act=None, bias_attr=None)
544 545 546
            proj_out, _ = fluid.layers.dynamic_lstmp(input=fc_out,
                                                     size=hidden_dim * 4,
                                                     proj_size=proj_dim,
Y
Yibing Liu 已提交
547 548 549 550
                                                     use_peepholes=False,
                                                     is_reverse=True,
                                                     cell_activation="tanh",
                                                     proj_activation="tanh")
Y
Yibing Liu 已提交
551
    """
552

Y
Yibing Liu 已提交
553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598
    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
599 600 601 602 603 604 605 606 607 608 609
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**

610
    Refer to `Empirical Evaluation of Gated Recurrent Neural Networks on
G
guosheng 已提交
611
    Sequence Modeling <https://arxiv.org/abs/1412.3555>`_
612

G
guosheng 已提交
613 614 615 616 617 618 619 620 621
    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)
622

G
guosheng 已提交
623
        h_t & = (1-u_t) \odot h_{t-1} + u_t \odot \\tilde{h_t}
624

G
guosheng 已提交
625
    The :math:`\odot` is the element-wise product of the vectors. :math:`act_g`
626 627
    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
628 629 630 631
    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
632
    to use fully-connect layer before GRU layer.
G
guosheng 已提交
633 634

    Args:
635 636
        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
637
            Variable is a matrix with shape :math:`(T \\times 3D)`, where
638
            :math:`T` is the total time steps in this mini-batch, :math:`D`
G
guosheng 已提交
639 640
            is the hidden size.
        size(int): The dimension of the gru cell.
641
        param_attr(ParamAttr|None): The parameter attribute for the learnable
G
guosheng 已提交
642 643
            hidden-hidden weight matrix. Note:

644
            - The shape of the weight matrix is :math:`(T \\times 3D)`, where
G
guosheng 已提交
645
              :math:`D` is the hidden size.
646
            - All elements in the weight matrix can be divided into two parts.
G
guosheng 已提交
647
              The first part are weights of the update gate and reset gate with
648
              shape :math:`(D \\times 2D)`, and the second part are weights for
G
guosheng 已提交
649
              candidate hidden state with shape :math:`(D \\times D)`.
650
        bias_attr(ParamAttr): The parameter attribute for learnable the
G
guosheng 已提交
651
            hidden-hidden bias.
652
        is_reverse(bool): Whether to compute reversed GRU, default
G
guosheng 已提交
653 654 655
            :attr:`False`.
        gate_activation(str): The activation for update gate and reset gate.
            Choices = ["sigmoid", "tanh", "relu", "identity"], default "sigmoid".
656
        activation(str): The activation for candidate hidden state.
G
guosheng 已提交
657 658 659
            Choices = ["sigmoid", "tanh", "relu", "identity"], default "tanh".

    Returns:
G
guosheng 已提交
660 661
        Variable: The hidden state of GRU. The shape is :math:`(T \\times D)`, \
            and lod is the same with the input.
662

G
guosheng 已提交
663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705
    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)
    inputs = {'Input': input, 'Weight': weight, 'Bias': bias}
    if h_0 != None:
        assert h_0.shape == (
            size, size), 'The shape of h0 should be(%d, %d)' % (size, size)
        inputs['h0'] = h_0

    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
706 707 708
def gru_unit(input,
             hidden,
             size,
709 710
             param_attr=None,
             bias_attr=None,
Y
Yu Yang 已提交
711
             activation='tanh',
712
             gate_activation='sigmoid'):
Y
Yu Yang 已提交
713
    """
714
    GRU unit layer. The equation of a gru step is:
Y
Yu Yang 已提交
715

716 717
        .. math::
            u_t & = actGate(xu_{t} + W_u h_{t-1} + b_u)
Y
Yu Yang 已提交
718

719
            r_t & = actGate(xr_{t} + W_r h_{t-1} + b_r)
Y
Yu Yang 已提交
720

721
            m_t & = actNode(xm_t + W_c dot(r_t, h_{t-1}) + b_m)
722

723
            h_t & = dot((1-u_t), m_t) + dot(u_t, h_{t-1})
724 725

    The inputs of gru unit includes :math:`z_t`, :math:`h_{t-1}`. In terms
726 727 728
    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
729 730
    connected layer has to be applied, such that :math:`z_t = W_{fc}x_t`.

731 732
    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
733 734 735
    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`.
736 737 738 739 740

    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.
741 742
        param_attr (ParamAttr): The weight parameters for gru unit. Default: None
        bias_attr (ParamAttr): The bias parameters for gru unit. Default: None
743 744 745 746
        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
747

748 749 750 751 752 753
    Returns:
        tuple: The hidden value, reset-hidden value and gate values.

    Examples:

        .. code-block:: python
Y
Yu Yang 已提交
754

755
             # assuming we have x_t_data and prev_hidden of size=10
756
             x_t = fluid.layers.fc(input=x_t_data, size=30)
757 758
             hidden_val, r_h_val, gate_val = fluid.layers.gru_unit(input=x_t,
                                                    hidden = prev_hidden)
Y
Yu Yang 已提交
759 760 761 762 763 764 765 766 767 768 769 770 771 772 773

    """
    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
774 775
    weight = helper.create_parameter(
        attr=helper.param_attr, shape=[size, 3 * size], dtype=dtype)
Y
Yu Yang 已提交
776

777 778 779 780
    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
781
    # create bias
782
    if helper.bias_attr:
Y
Yu Yang 已提交
783 784 785
        bias_size = [1, 3 * size]
        bias = helper.create_parameter(
            attr=helper.bias_attr, shape=bias_size, dtype=dtype, is_bias=True)
786
        inputs['Bias'] = bias
Y
Yu Yang 已提交
787 788 789

    helper.append_op(
        type='gru_unit',
790
        inputs=inputs,
Y
Yu Yang 已提交
791 792 793 794 795 796
        outputs={
            'Gate': gate,
            'ResetHiddenPrev': reset_hidden_pre,
            'Hidden': updated_hidden,
        },
        attrs={
797 798
            'activation': 2,  # tanh
            'gate_activation': 1,  # sigmoid
Y
Yu Yang 已提交
799 800 801 802 803
        })

    return updated_hidden, reset_hidden_pre, gate


804
def linear_chain_crf(input, label, param_attr=None):
Y
Yu Yang 已提交
805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829
    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


830
def crf_decoding(input, param_attr, label=None):
Y
Yu Yang 已提交
831 832 833 834 835 836 837 838 839 840 841 842 843
    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
844
def cos_sim(X, Y):
Y
Yu Yang 已提交
845 846 847 848
    """
    This function performs the cosine similarity between two tensors
    X and Y and returns that as the output.
    """
F
fengjiayi 已提交
849
    helper = LayerHelper('cos_sim', **locals())
Y
Yu Yang 已提交
850 851 852 853 854 855 856 857 858 859 860 861 862
    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


863
def dropout(x, dropout_prob, is_test=False, seed=None, name=None):
864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880
    """
    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:
       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.
881 882
       name(str|None): A name for this layer(optional). If set None, the layer
                    will be named automatically.
883 884 885 886 887 888 889 890 891 892 893

    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
894
    helper = LayerHelper('dropout', **locals())
895 896 897 898 899 900 901
    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]},
902 903 904 905 906 907
        attrs={
            'dropout_prob': dropout_prob,
            'is_test': is_test,
            'fix_seed': seed is not None,
            'seed': seed if seed is not None else 0
        })
908 909 910
    return out


F
fengjiayi 已提交
911
def cross_entropy(input, label, soft_label=False):
Y
Yu Yang 已提交
912
    """
Y
Yibing Liu 已提交
913 914
    **Cross Entropy Layer**

915 916 917
    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
918 919

    1) One-hot cross-entropy:
F
fengjiayi 已提交
920
        `soft_label = False`, `Label[i, 0]` indicates the class index for sample i:
Y
yangyaming 已提交
921

Y
Yibing Liu 已提交
922
        .. math::
Y
yangyaming 已提交
923

Y
Yibing Liu 已提交
924 925 926
            Y[i] = -\log(X[i, Label[i]])

    2) Soft-label cross-entropy:
F
fengjiayi 已提交
927 928
        `soft_label = True`, `Label[i, j]` indicates the soft label of class j
        for sample i:
Y
Yibing Liu 已提交
929 930 931 932 933

        .. math::

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

Y
Yibing Liu 已提交
934
       Please make sure that in this case the summation of each row of `label`
Y
Yibing Liu 已提交
935 936 937
       equals one.

    3) One-hot cross-entropy with vecterized `label`:
F
fengjiayi 已提交
938 939
         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
940
         to a one-hot cross-entropy with one-hot label representation.
Y
yangyaming 已提交
941

Y
Yibing Liu 已提交
942
    Args:
Y
yangyaming 已提交
943
        input (Variable|list):  a 2-D tensor with shape [N x D], where N is the
944 945 946 947
                                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
948
        label (Variable|list): the ground truth which is a 2-D tensor. When
949 950 951 952
                               `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
953
        soft_label (bool): a flag indicating whether to
954 955
                                           interpretate the given labels as soft
                                           labels, default `False`.
Y
Yibing Liu 已提交
956 957 958 959 960

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

    Raises:
961 962 963 964 965
        `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
966 967 968 969 970 971

    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
972
    """
F
fengjiayi 已提交
973
    helper = LayerHelper('cross_entropy', **locals())
Y
Yu Yang 已提交
974 975 976 977 978 979
    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
980
        attrs={"soft_label": soft_label})
Y
Yu Yang 已提交
981 982 983
    return out


F
fengjiayi 已提交
984
def square_error_cost(input, label):
Y
Yu Yang 已提交
985
    """
986 987
    **Square error cost layer**

988 989
    This layer accepts input predictions and target label and returns the
    squared error cost.
Y
ying 已提交
990

991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007
    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:
       input(Variable): Input tensor, has predictions.
       label(Variable): Label tensor, has target labels.

    Returns:
G
guosheng 已提交
1008
        Variable: The tensor variable storing the element-wise squared error \
1009
                  difference of input and label.
1010 1011 1012 1013 1014 1015 1016 1017

    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
1018
    """
F
fengjiayi 已提交
1019
    helper = LayerHelper('square_error_cost', **locals())
Y
Yu Yang 已提交
1020 1021 1022 1023 1024 1025 1026 1027 1028
    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
1029 1030
        type='square', inputs={'X': [minus_out]},
        outputs={'Out': [square_out]})
Y
Yu Yang 已提交
1031 1032 1033 1034 1035 1036 1037
    return square_out


def chunk_eval(input,
               label,
               chunk_scheme,
               num_chunk_types,
F
fengjiayi 已提交
1038
               excluded_chunk_types=None):
Y
Yu Yang 已提交
1039
    """
Y
yangyaming 已提交
1040
    This function computes and outputs the precision, recall and
1041
    F1-score of chunk detection.
Y
Yu Yang 已提交
1042
    """
F
fengjiayi 已提交
1043
    helper = LayerHelper("chunk_eval", **locals())
Y
Yu Yang 已提交
1044 1045 1046 1047 1048

    # prepare output
    precision = helper.create_tmp_variable(dtype="float32")
    recall = helper.create_tmp_variable(dtype="float32")
    f1_score = helper.create_tmp_variable(dtype="float32")
1049 1050 1051
    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
1052 1053 1054 1055 1056 1057 1058 1059

    helper.append_op(
        type="chunk_eval",
        inputs={"Inference": [input],
                "Label": [label]},
        outputs={
            "Precision": [precision],
            "Recall": [recall],
1060 1061 1062 1063
            "F1-Score": [f1_score],
            "NumInferChunks": [num_infer_chunks],
            "NumLabelChunks": [num_label_chunks],
            "NumCorrectChunks": [num_correct_chunks]
Y
Yu Yang 已提交
1064 1065 1066
        },
        attrs={
            "num_chunk_types": num_chunk_types,
G
guosheng 已提交
1067 1068
            "chunk_scheme": chunk_scheme,
            "excluded_chunk_types": excluded_chunk_types or []
Y
Yu Yang 已提交
1069
        })
1070 1071
    return (precision, recall, f1_score, num_infer_chunks, num_label_chunks,
            num_correct_chunks)
Y
Yu Yang 已提交
1072 1073 1074 1075 1076 1077 1078 1079 1080


def sequence_conv(input,
                  num_filters,
                  filter_size=3,
                  filter_stride=1,
                  padding=None,
                  bias_attr=None,
                  param_attr=None,
1081
                  act=None):
Y
Yu Yang 已提交
1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114
    """
    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.
    """

    # FIXME(dzh) : want to unify the argument of python layer
    # function. So we ignore some unecessary attributes.
    # such as, padding_trainable, context_start.

    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)


1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126
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


1127
def softmax(input, param_attr=None, bias_attr=None, use_cudnn=True, name=None):
1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138
    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
1139 1140 1141
def conv2d(input,
           num_filters,
           filter_size,
C
chengduoZH 已提交
1142 1143
           stride=1,
           padding=0,
1144
           dilation=1,
Y
Yu Yang 已提交
1145 1146 1147
           groups=None,
           param_attr=None,
           bias_attr=None,
C
chengduoZH 已提交
1148
           use_cudnn=True,
1149
           use_mkldnn=False,
1150 1151
           act=None,
           name=None):
Y
Yu Yang 已提交
1152
    """
C
chengduoZH 已提交
1153 1154 1155
    **Convlution2D Layer**

    The convolution2D layer calculates the output based on the input, filter
1156 1157 1158
    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
1159 1160
    The details of convolution layer, please refer UFLDL's `convolution,
    <http://ufldl.stanford.edu/tutorial/supervised/FeatureExtractionUsingConvolution/>`_ .
1161 1162 1163
    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
1164

1165
    For each input :math:`X`, the equation is:
C
refine  
chengduoZH 已提交
1166

C
chengduoZH 已提交
1167 1168
    .. math::

C
refine  
chengduoZH 已提交
1169
        Out = \sigma (W \\ast X + b)
C
chengduoZH 已提交
1170

C
chengduoZH 已提交
1171
    In the above equation:
C
chengduoZH 已提交
1172

1173 1174 1175 1176 1177
    * :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.
1178 1179
    * :math:`Out`: Output value, the shape of :math:`Out` and :math:`X` may be
                   different.
C
chengduoZH 已提交
1180 1181 1182

    Example:

1183 1184 1185
        - Input:

          Input shape: $(N, C_{in}, H_{in}, W_{in})$
C
refine  
chengduoZH 已提交
1186

1187
          Filter shape: $(C_{out}, C_{in}, H_f, W_f)$
C
refine  
chengduoZH 已提交
1188

1189 1190
        - Output:
          Output shape: $(N, C_{out}, H_{out}, W_{out})$
C
refine  
chengduoZH 已提交
1191

C
chengduoZH 已提交
1192
        Where
1193 1194

        .. math::
C
chengduoZH 已提交
1195

C
chengduoZH 已提交
1196 1197
        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
1198 1199

    Args:
1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211
       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.
1212 1213 1214
       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.
1215 1216 1217 1218 1219 1220 1221 1222 1223 1224
       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
       act(str): Activation type. Default: None
1225 1226
       name(str|None): A name for this layer(optional). If set None, the layer
           will be named automatically.
C
chengduoZH 已提交
1227 1228

    Returns:
G
guosheng 已提交
1229
        Variable: The tensor variable storing the convolution and \
C
chengduoZH 已提交
1230 1231
                  non-linearity activation result.

C
refine  
chengduoZH 已提交
1232
    Raises:
1233 1234
        ValueError: If the shapes of input, filter_size, stride, padding and
                    groups mismatch.
C
refine  
chengduoZH 已提交
1235

C
chengduoZH 已提交
1236 1237 1238
    Examples:
        .. code-block:: python

1239 1240 1241 1242
          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
1243 1244 1245 1246 1247
    """
    if stride is None:
        stride = [1, 1]

    num_channels = input.shape[1]
1248 1249

    l_type = 'conv2d'
X
xzl 已提交
1250 1251
    if (num_channels == groups and num_filters % num_channels == 0 and
            not use_cudnn):
1252
        l_type = 'depthwise_conv2d'
1253 1254 1255 1256

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

Y
Yu Yang 已提交
1257 1258 1259 1260 1261 1262 1263
    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
1264 1265 1266
    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')
1267
    dilation = utils.convert_to_list(dilation, 2, 'dilation')
C
chengduoZH 已提交
1268

C
chengduoZH 已提交
1269 1270
    if not isinstance(use_cudnn, bool):
        raise ValueError("use_cudnn should be True or False")
Y
Yu Yang 已提交
1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287

    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(
1288
        type=l_type,
Y
Yu Yang 已提交
1289 1290 1291 1292 1293
        inputs={
            'Input': input,
            'Filter': filter_param,
        },
        outputs={"Output": pre_bias},
C
chengduoZH 已提交
1294 1295 1296
        attrs={
            'strides': stride,
            'paddings': padding,
1297
            'dilations': dilation,
C
chengduoZH 已提交
1298
            'groups': groups,
1299 1300
            'use_cudnn': use_cudnn,
            'use_mkldnn': use_mkldnn
C
chengduoZH 已提交
1301
        })
Y
Yu Yang 已提交
1302 1303 1304 1305 1306 1307

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

    return helper.append_activation(pre_act)


F
fengjiayi 已提交
1308
def sequence_pool(input, pool_type):
Y
Yu Yang 已提交
1309
    """
Y
yangyaming 已提交
1310 1311 1312
    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
1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337

    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)
1338 1339
         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
1340

L
Luo Tao 已提交
1341 1342
    Args:
        input(variable): The input variable which is a LoDTensor.
Y
yangyaming 已提交
1343
        pool_type (string): The pooling type of sequence_pool.
L
Luo Tao 已提交
1344 1345 1346 1347 1348 1349 1350 1351
            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
1352

Y
yangyaming 已提交
1353
             x = fluid.layers.data(name='x', shape=[7, 1],
L
Luo Tao 已提交
1354 1355 1356 1357 1358
                              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')
1359 1360
             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
1361
    """
F
fengjiayi 已提交
1362
    helper = LayerHelper('sequence_pool', **locals())
Y
Yu Yang 已提交
1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373
    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
1374 1375 1376 1377 1378
    # 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
1379 1380 1381
    return pool_out


F
fengjiayi 已提交
1382
def sequence_first_step(input):
L
Luo Tao 已提交
1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396
    """
    This funciton get the first step of sequence.

    .. code-block:: text

       x is a 1-level LoDTensor:
         x.lod = [[0, 2, 5, 7]]
         x.data = [1, 3, 2, 4, 6, 5, 1]
         x.dims = [7, 1]

       then output is a Tensor:
         out.dim = [3, 1]
         with condition len(x.lod[-1]) - 1 == out.dims[0]
         out.data = [1, 2, 5], where 1=first(1,3), 2=first(2,4,6), 5=first(5,1)
F
fengjiayi 已提交
1397

L
Luo Tao 已提交
1398 1399 1400 1401 1402 1403 1404 1405 1406
    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
1407

Y
yangyaming 已提交
1408
             x = fluid.layers.data(name='x', shape=[7, 1],
L
Luo Tao 已提交
1409 1410 1411
                              dtype='float32', lod_level=1)
             x_first_step = fluid.layers.sequence_first_step(input=x)
    """
1412 1413 1414
    return sequence_pool(input=input, pool_type="first")


F
fengjiayi 已提交
1415
def sequence_last_step(input):
L
Luo Tao 已提交
1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429
    """
    This funciton get the last step of sequence.

    .. code-block:: text

       x is a 1-level LoDTensor:
         x.lod = [[0, 2, 5, 7]]
         x.data = [1, 3, 2, 4, 6, 5, 1]
         x.dims = [7, 1]

       then output is a Tensor:
         out.dim = [3, 1]
         with condition len(x.lod[-1]) - 1 == out.dims[0]
         out.data = [3, 6, 1], where 3=last(1,3), 6=last(2,4,6), 1=last(5,1)
F
fengjiayi 已提交
1430

L
Luo Tao 已提交
1431 1432 1433 1434 1435 1436 1437 1438 1439
    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
1440

Y
yangyaming 已提交
1441
             x = fluid.layers.data(name='x', shape=[7, 1],
L
Luo Tao 已提交
1442 1443 1444
                              dtype='float32', lod_level=1)
             x_last_step = fluid.layers.sequence_last_step(input=x)
    """
1445 1446 1447
    return sequence_pool(input=input, pool_type="last")


Y
Yu Yang 已提交
1448
def pool2d(input,
C
chengduoZH 已提交
1449 1450
           pool_size=-1,
           pool_type="max",
C
chengduoZH 已提交
1451 1452
           pool_stride=1,
           pool_padding=0,
C
caoying03 已提交
1453
           global_pooling=False,
C
chengduoZH 已提交
1454
           use_cudnn=True,
1455
           ceil_mode=False,
1456
           use_mkldnn=False,
C
caoying03 已提交
1457
           name=None):
Y
Yu Yang 已提交
1458 1459 1460 1461 1462 1463 1464 1465
    """
    This function adds the operator for pooling in 2 dimensions, using the
    pooling configurations mentioned in input parameters.
    """
    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
1466

C
chengduoZH 已提交
1467 1468 1469 1470 1471
    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
1472 1473 1474 1475
    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
1476 1477
    if not isinstance(use_cudnn, bool):
        raise ValueError("use_cudnn should be True or False")
Y
Yu Yang 已提交
1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491

    helper = LayerHelper('pool2d', **locals())
    dtype = helper.input_dtype()
    pool_out = helper.create_tmp_variable(dtype)

    helper.append_op(
        type="pool2d",
        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
1492
            "paddings": pool_padding,
1493
            "use_cudnn": use_cudnn,
1494 1495
            "ceil_mode": ceil_mode,
            "use_mkldnn": use_mkldnn
Y
Yu Yang 已提交
1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507
        })

    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
1508
               data_layout='NCHW',
Y
Yang Yang 已提交
1509
               in_place=False,
1510
               use_mkldnn=False,
1511 1512
               name=None,
               moving_mean_name=None,
W
wanghaoshuang 已提交
1513
               moving_variance_name=None,
W
wanghaoshuang 已提交
1514
               do_model_average_for_mean_and_var=False):
Y
Yu Yang 已提交
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
    """
    This function helps create an operator to implement
    the BatchNorm layer using the configurations from the input parameters.
    """
    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(
1541
        attr=helper.bias_attr, shape=param_shape, dtype=dtype, is_bias=True)
Y
Yu Yang 已提交
1542

1543 1544
    mean = helper.create_parameter(
        attr=ParamAttr(
W
wanghaoshuang 已提交
1545 1546 1547
            name=moving_mean_name,
            initializer=Constant(0.0),
            trainable=False,
W
wanghaoshuang 已提交
1548
            do_model_average=do_model_average_for_mean_and_var),
Q
QI JUN 已提交
1549
        shape=param_shape,
1550 1551 1552 1553 1554 1555 1556
        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
1557
            trainable=False,
W
wanghaoshuang 已提交
1558
            do_model_average=do_model_average_for_mean_and_var),
Q
QI JUN 已提交
1559
        shape=param_shape,
1560 1561
        dtype=input.dtype)
    variance.stop_gradient = True
Y
Yu Yang 已提交
1562 1563 1564 1565 1566 1567

    # 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
1568 1569
    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
1570

Y
Yang Yang 已提交
1571
    batch_norm_out = input if in_place else helper.create_tmp_variable(dtype)
Y
Yu Yang 已提交
1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588

    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
        },
1589 1590 1591 1592 1593 1594
        attrs={
            "momentum": momentum,
            "epsilon": epsilon,
            "is_test": is_test,
            "use_mkldnn": use_mkldnn
        })
Y
Yu Yang 已提交
1595 1596 1597 1598

    return helper.append_activation(batch_norm_out)


G
guosheng 已提交
1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610
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**

1611
    Assume feature vectors exist on dimensions
G
guosheng 已提交
1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631
    :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.
1632
        scale(bool): Whether to learn the adaptive gain :math:`g` after
G
guosheng 已提交
1633
            normalization.
1634
        shift(bool): Whether to learn the adaptive bias :math:`b` after
G
guosheng 已提交
1635
            normalization.
1636
        begin_norm_axis(bool): The normalization will be performed along
G
guosheng 已提交
1637
            dimensions from :attr:`begin_norm_axis` to :attr:`rank(input)`.
1638
        epsilon(float): The small value added to the variance to prevent
G
guosheng 已提交
1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669
            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.

    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
1670
    if shift:
G
guosheng 已提交
1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694
        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
1695
def beam_search_decode(ids, scores, name=None):
Y
Yu Yang 已提交
1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715
    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
1716 1717 1718
                     padding=0,
                     stride=1,
                     dilation=1,
1719
                     groups=None,
C
caoying03 已提交
1720
                     param_attr=None,
1721
                     bias_attr=None,
C
chengduoZH 已提交
1722
                     use_cudnn=True,
1723
                     act=None,
C
caoying03 已提交
1724
                     name=None):
Y
Yu Yang 已提交
1725
    """
1726 1727 1728 1729 1730 1731 1732 1733
    **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
1734 1735
    layer, please refer to the following explanation and references
    `therein <http://www.matthewzeiler.com/wp-content/uploads/2017/07/cvpr2010.pdf>`_.
1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747

    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.
1748 1749
    * :math:`Out`: Output value, the shape of :math:`Out` and :math:`X` may be
                   different.
Y
Yu Yang 已提交
1750

1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763
    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
1764

1765 1766 1767 1768
        .. 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
1769 1770

    Args:
1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789
       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.
1790 1791 1792 1793 1794 1795
       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
1796 1797
       param_attr(ParamAttr): The parameters to the Conv2d_transpose Layer.
                              Default: None
1798
       bias_attr(ParamAttr): Bias parameter for the Conv2d layer. Default: None
1799 1800
       use_cudnn(bool): Use cudnn kernel or not, it is valid only when the cudnn
           library is installed. Default: True
1801
       act(str): Activation type. Default: None
1802 1803
       name(str|None): A name for this layer(optional). If set None, the layer
           will be named automatically.
Y
Yu Yang 已提交
1804 1805

    Returns:
1806 1807 1808
       Variable: The tensor variable storing the convolution transpose result.

    Raises:
1809 1810
       ValueError: If the shapes of input, filter_size, stride, padding and
                   groups mismatch.
1811 1812 1813 1814

    Examples:
       .. code-block:: python

1815 1816 1817 1818
          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
1819 1820 1821 1822 1823 1824
    """
    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
1825 1826 1827
    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
1828

C
chengduoZH 已提交
1829 1830 1831
    if not isinstance(use_cudnn, bool):
        raise ValueError("use_cudnn should be True or False")

Y
Yu Yang 已提交
1832 1833 1834 1835 1836 1837 1838 1839
    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
1840 1841 1842 1843 1844

        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
1845
        filter_size = [filter_size_h, filter_size_w]
C
chengduoZH 已提交
1846 1847 1848
    else:
        filter_size = utils.convert_to_list(filter_size, 2,
                                            'conv2d_transpose.filter_size')
Y
Yu Yang 已提交
1849

1850 1851
    groups = 1 if groups is None else groups
    filter_shape = [input_channel, num_filters / groups] + filter_size
Y
Yu Yang 已提交
1852 1853 1854
    img_filter = helper.create_parameter(
        dtype=input.dtype, shape=filter_shape, attr=helper.param_attr)

1855
    pre_bias = helper.create_tmp_variable(dtype=input.dtype)
Y
Yu Yang 已提交
1856 1857 1858 1859
    helper.append_op(
        type='conv2d_transpose',
        inputs={'Input': [input],
                'Filter': [img_filter]},
1860
        outputs={'Output': pre_bias},
C
chengduoZH 已提交
1861 1862 1863 1864
        attrs={
            'strides': stride,
            'paddings': padding,
            'dilations': dilation,
1865
            'groups': groups,
C
chengduoZH 已提交
1866 1867
            'use_cudnn': use_cudnn
        })
Y
Yu Yang 已提交
1868

1869 1870
    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
1871
    return out
Y
yangyaming 已提交
1872 1873


Y
yangyaming 已提交
1874
def sequence_expand(x, y, ref_level=-1, name=None):
1875
    """Sequence Expand Layer. This layer will expand the input variable **x**
Y
yangyaming 已提交
1876 1877 1878 1879
    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:
1880 1881 1882 1883 1884

    .. code-block:: text

        * Case 1
            x is a LoDTensor:
Y
yangyaming 已提交
1885 1886
                x.lod  = [[0,   2,        4]]
                x.data = [[a], [b], [c], [d]]
1887 1888 1889 1890 1891 1892
                x.dims = [4, 1]

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

Y
yangyaming 已提交
1893
            ref_level: 0
1894

Y
yangyaming 已提交
1895 1896 1897
            then output is a 1-level LoDTensor:
                out.lod =  [[0,   2,        4,        6,        8]]
                out.data = [[a], [b], [a], [b], [c], [d], [c], [d]]
1898 1899 1900 1901
                out.dims = [8, 1]

        * Case 2
            x is a Tensor:
Y
yangyaming 已提交
1902
                x.data = [[a], [b], [c]]
1903 1904 1905
                x.dims = [3, 1]

            y is a LoDTensor:
Y
yangyaming 已提交
1906
                y.lod = [[0, 2, 2, 5]]
1907

Y
yangyaming 已提交
1908
            ref_level: -1
1909

Y
yangyaming 已提交
1910 1911 1912
            then output is a Tensor:
                out.data = [[a], [a], [c], [c], [c]]
                out.dims = [5, 1]
1913 1914 1915
    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
1916 1917
        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
1918
        name(str|None): A name for this layer(optional). If set None, the layer
Y
yangyaming 已提交
1919
                        will be named automatically.
1920 1921 1922 1923 1924 1925 1926 1927 1928 1929

    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
1930
            out = layers.sequence_expand(x=x, y=y, ref_level=0)
1931
    """
Y
yangyaming 已提交
1932
    helper = LayerHelper('sequence_expand', input=x, **locals())
1933 1934 1935
    dtype = helper.input_dtype()
    tmp = helper.create_tmp_variable(dtype)
    helper.append_op(
Y
yangyaming 已提交
1936 1937 1938 1939 1940
        type='sequence_expand',
        inputs={'X': x,
                'Y': y},
        outputs={'Out': tmp},
        attrs={'ref_level': ref_level})
1941
    return tmp
1942 1943


Q
Qiao Longfei 已提交
1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975
def beam_search(pre_ids, ids, scores, beam_size, end_id, level=0):
    '''
    This function implements the beam search algorithm.
    '''
    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
1976 1977 1978 1979
def lstm_unit(x_t,
              hidden_t_prev,
              cell_t_prev,
              forget_bias=0.0,
Y
yangyaming 已提交
1980
              param_attr=None,
C
caoying03 已提交
1981 1982
              bias_attr=None,
              name=None):
Y
yangyaming 已提交
1983 1984 1985 1986
    """Lstm unit layer. The equation of a lstm step is:

        .. math::

1987
            i_t & = \sigma(W_{x_i}x_{t} + W_{h_i}h_{t-1} + b_i)
Y
yangyaming 已提交
1988

1989
            f_t & = \sigma(W_{x_f}x_{t} + W_{h_f}h_{t-1} + b_f)
Y
yangyaming 已提交
1990

1991
            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
1992

1993
            o_t & = \sigma(W_{x_o}x_{t} + W_{h_o}h_{t-1} + b_o)
Y
yangyaming 已提交
1994 1995 1996

            h_t & = o_t tanh(c_t)

1997 1998 1999 2000 2001 2002
    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
2003 2004 2005

        .. math::

2006
            L_{i_t} = W_{x_i}x_{t} + W_{h_i}h_{t-1} + b_i
Y
yangyaming 已提交
2007 2008 2009 2010 2011 2012 2013 2014

    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
2015
    This layer has two outputs including :math:`h_t` and :math:`o_t`.
Y
yangyaming 已提交
2016 2017

    Args:
Y
yangyaming 已提交
2018 2019 2020 2021 2022 2023
        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
2024
        forget_bias (float): The forget bias of lstm unit.
Y
yangyaming 已提交
2025 2026
        param_attr (ParamAttr): The attributes of parameter weights, used to set
            initializer, name etc.
Y
yangyaming 已提交
2027 2028
        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
2029 2030
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
Y
yangyaming 已提交
2031 2032

    Returns:
Y
yangyaming 已提交
2033
        tuple: The hidden value and cell value of lstm unit.
Y
yangyaming 已提交
2034 2035

    Raises:
2036 2037 2038 2039
        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
2040 2041 2042 2043 2044 2045

    Examples:

        .. code-block:: python

             x_t = fluid.layers.fc(input=x_t_data, size=10)
2046
             prev_hidden = fluid.layers.fc(input=prev_hidden_data, size=30)
Y
yangyaming 已提交
2047
             prev_cell = fluid.layers.fc(input=prev_cell_data, size=30)
Y
yangyaming 已提交
2048
             hidden_value, cell_value = fluid.layers.lstm_unit(x_t=x_t,
Y
yangyaming 已提交
2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064
                                                    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
2065
        raise ValueError("The 1st dimensions of x_t, hidden_t_prev and "
2066 2067 2068 2069
                         "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
2070 2071
                         "cell_t_prev must be the same.")

Y
yangyaming 已提交
2072 2073 2074
    if bias_attr is None:
        bias_attr = ParamAttr()

Y
yangyaming 已提交
2075
    size = cell_t_prev.shape[1]
2076
    concat_out = concat(input=[x_t, hidden_t_prev], axis=1)
Y
yangyaming 已提交
2077 2078
    fc_out = fc(input=concat_out,
                size=4 * size,
Y
yangyaming 已提交
2079
                param_attr=param_attr,
2080
                bias_attr=bias_attr)
Y
yangyaming 已提交
2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092
    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
2093
    return h, c
G
guosheng 已提交
2094 2095


C
caoying03 已提交
2096
def reduce_sum(input, dim=None, keep_dim=False, name=None):
G
guosheng 已提交
2097
    """
Y
yangyaming 已提交
2098
    Computes the sum of tensor elements over the given dimension.
G
guosheng 已提交
2099 2100 2101

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
W
whs 已提交
2102
        dim (list|int|None): The dimensions along which the sum is performed. If
Y
yangyaming 已提交
2103 2104
            :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
2105 2106
            range :math:`[-rank(input), rank(input))`. If :math:`dim[i] < 0`,
            the dimension to reduce is :math:`rank + dim[i]`.
2107
        keep_dim (bool|False): Whether to reserve the reduced dimension in the
Y
yangyaming 已提交
2108
            output Tensor. The result tensor will have one fewer dimension
G
guosheng 已提交
2109
            than the :attr:`input` unless :attr:`keep_dim` is true.
C
caoying03 已提交
2110 2111
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
G
guosheng 已提交
2112 2113 2114

    Returns:
        Variable: The reduced Tensor variable.
F
fengjiayi 已提交
2115

G
guosheng 已提交
2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126
    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
2127 2128 2129 2130 2131 2132 2133 2134

            # 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
2135 2136 2137
    """
    helper = LayerHelper('reduce_sum', **locals())
    out = helper.create_tmp_variable(dtype=helper.input_dtype())
W
whs 已提交
2138 2139
    if dim is not None and not isinstance(dim, list):
        dim = [dim]
G
guosheng 已提交
2140 2141 2142 2143 2144
    helper.append_op(
        type='reduce_sum',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
W
whs 已提交
2145
            'dim': dim if dim != None else [0],
G
guosheng 已提交
2146 2147 2148 2149
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out
G
guosheng 已提交
2150 2151


C
caoying03 已提交
2152
def reduce_mean(input, dim=None, keep_dim=False, name=None):
G
guosheng 已提交
2153
    """
Y
yangyaming 已提交
2154
    Computes the mean of tensor elements over the given dimension.
G
guosheng 已提交
2155 2156 2157

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
W
whs 已提交
2158
        dim (list|int|None): The dimensions along which the mean is computed. If
Y
yangyaming 已提交
2159 2160 2161
            :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
2162
            :math:`dim[i] < 0`, the dimension to reduce is :math:`rank + dim[i]`.
Y
yangyaming 已提交
2163 2164
        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
2165
            than the :attr:`input` unless :attr:`keep_dim` is true.
C
caoying03 已提交
2166 2167
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
G
guosheng 已提交
2168 2169 2170

    Returns:
        Variable: The reduced Tensor variable.
F
fengjiayi 已提交
2171

G
guosheng 已提交
2172 2173 2174 2175 2176 2177 2178 2179 2180 2181
    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
2182 2183
            fluid.layers.reduce_mean(
                x, dim=1, keep_dim=True)  # [[0.475], [0.4]]
W
whs 已提交
2184 2185 2186 2187 2188 2189 2190

            # 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
2191 2192 2193
    """
    helper = LayerHelper('reduce_mean', **locals())
    out = helper.create_tmp_variable(dtype=helper.input_dtype())
W
whs 已提交
2194 2195
    if dim is not None and not isinstance(dim, list):
        dim = [dim]
G
guosheng 已提交
2196 2197 2198 2199 2200
    helper.append_op(
        type='reduce_mean',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
W
whs 已提交
2201
            'dim': dim if dim != None else [0],
G
guosheng 已提交
2202 2203 2204 2205
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out
2206 2207


C
caoying03 已提交
2208
def reduce_max(input, dim=None, keep_dim=False, name=None):
2209
    """
Y
yangyaming 已提交
2210
    Computes the maximum of tensor elements over the given dimension.
2211 2212 2213

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
W
whs 已提交
2214
        dim (list|int|None): The dimension along which the maximum is computed.
Y
yangyaming 已提交
2215 2216 2217
            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
2218
            If :math:`dim[i] < 0`, the dimension to reduce is :math:`rank + dim[i]`.
Y
yangyaming 已提交
2219 2220
        keep_dim (bool): Whether to reserve the reduced dimension in the
            output Tensor. The result tensor will have one fewer dimension
2221
            than the :attr:`input` unless :attr:`keep_dim` is true.
C
caoying03 已提交
2222 2223
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
2224 2225 2226

    Returns:
        Variable: The reduced Tensor variable.
Y
yangyaming 已提交
2227

2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238
    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
2239 2240 2241 2242 2243 2244 2245

            # 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]
2246 2247 2248
    """
    helper = LayerHelper('reduce_max', **locals())
    out = helper.create_tmp_variable(dtype=helper.input_dtype())
W
whs 已提交
2249 2250
    if dim is not None and not isinstance(dim, list):
        dim = [dim]
2251 2252 2253 2254 2255
    helper.append_op(
        type='reduce_max',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
W
whs 已提交
2256
            'dim': dim if dim != None else [0],
2257 2258 2259 2260 2261 2262
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out


C
caoying03 已提交
2263
def reduce_min(input, dim=None, keep_dim=False, name=None):
2264
    """
Y
yangyaming 已提交
2265
    Computes the minimum of tensor elements over the given dimension.
2266 2267 2268

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
W
whs 已提交
2269
        dim (list|int|None): The dimensions along which the minimum is computed.
Y
yangyaming 已提交
2270 2271 2272
            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
2273
            If :math:`dim[i] < 0`, the dimension to reduce is :math:`rank + dim[i]`.
Y
yangyaming 已提交
2274 2275
        keep_dim (bool): Whether to reserve the reduced dimension in the
            output Tensor. The result tensor will have one fewer dimension
2276
            than the :attr:`input` unless :attr:`keep_dim` is true.
C
caoying03 已提交
2277 2278
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
2279 2280 2281

    Returns:
        Variable: The reduced Tensor variable.
Y
yangyaming 已提交
2282

2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293
    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
2294 2295 2296 2297 2298 2299 2300

            # 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]
2301 2302 2303
    """
    helper = LayerHelper('reduce_min', **locals())
    out = helper.create_tmp_variable(dtype=helper.input_dtype())
W
whs 已提交
2304 2305
    if dim is not None and not isinstance(dim, list):
        dim = [dim]
2306 2307 2308 2309 2310
    helper.append_op(
        type='reduce_min',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
W
whs 已提交
2311
            'dim': dim if dim != None else [0],
2312 2313 2314 2315
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out
G
guosheng 已提交
2316 2317


2318 2319 2320 2321 2322 2323
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
2324
        dim (list|int|None): The dimensions along which the product is performed. If
2325 2326
            :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
2327 2328
            range :math:`[-rank(input), rank(input))`. If :math:`dim[i] < 0`,
            the dimension to reduce is :math:`rank + dim[i]`.
2329 2330 2331
        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
2332
        name(str|None): A name for this layer(optional). If set None, the
Z
zhouhanqing 已提交
2333
            layer will be named automatically.
2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347

    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
2348
            fluid.layers.reduce_prod(x, dim=1,
Z
zhouhanqing 已提交
2349
                                     keep_dim=True)  # [[0.027], [0.0084]]
W
whs 已提交
2350 2351 2352 2353 2354 2355 2356

            # 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]
2357 2358 2359
    """
    helper = LayerHelper('reduce_prod', **locals())
    out = helper.create_tmp_variable(dtype=helper.input_dtype())
W
whs 已提交
2360 2361
    if dim is not None and not isinstance(dim, list):
        dim = [dim]
2362 2363 2364 2365 2366
    helper.append_op(
        type='reduce_prod',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
W
whs 已提交
2367
            'dim': dim if dim != None else [0],
2368 2369 2370 2371 2372 2373
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out


C
caoying03 已提交
2374
def split(input, num_or_sections, dim=-1, name=None):
G
guosheng 已提交
2375
    """
C
caoying03 已提交
2376
    Split the input tensor into multiple sub-tensors.
G
guosheng 已提交
2377 2378 2379

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
C
caoying03 已提交
2380 2381 2382 2383 2384
        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
2385
            :attr:`dim` dimension orderly.
C
caoying03 已提交
2386
        dim (int): The dimension along which to split. If :math:`dim < 0`, the
G
guosheng 已提交
2387
            dimension to split along is :math:`rank(input) + dim`.
C
caoying03 已提交
2388 2389
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
G
guosheng 已提交
2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401

    Returns:
        List: The list of segmented tensor variables.

    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
2402 2403
            x0, x1, x2 = fluid.layers.split(
                x, num_or_sections=[2, 3, 4], dim=1)
G
guosheng 已提交
2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432
            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
2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465


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

    output = x / sqrt(max(sum(x**2), epsilon))

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

    Args:
       x(Variable|list): The input tensor to l2_normalize layer.
       axis(int): Dimension along which to normalize the input.
       epsilon(float): A lower bound value for `x`'s l2 norm. sqrt(epsilon) will
                       be used as the divisor if the l2 norm of `x` is less than
                       sqrt(epsilon).
       name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.


    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
2466
          normed = fluid.layers.l2_normalize(x=data, axis=1)
C
caoying03 已提交
2467 2468
    """

F
fengjiayi 已提交
2469 2470
    if len(x.shape) == 1:
        axis = 0
C
caoying03 已提交
2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481
    helper = LayerHelper("l2_normalize", **locals())

    square = helper.create_tmp_variable(dtype=x.dtype)
    helper.append_op(type="square", inputs={"X": x}, outputs={"Out": square})

    reduced_sum = helper.create_tmp_variable(dtype=x.dtype)
    helper.append_op(
        type="reduce_sum",
        inputs={"X": square},
        outputs={"Out": reduced_sum},
        attrs={
W
whs 已提交
2482
            "dim": [1] if axis is None else [axis],
C
caoying03 已提交
2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495
            "keep_dim": True,
            "reduce_all": False
        })

    # TODO(caoying) A lower bound value epsilon for the norm is needed to
    # imporve the numeric stability of reciprocal. This requires a maximum_op.
    rsquare = helper.create_tmp_variable(dtype=x.dtype)
    helper.append_op(
        type="reciprocal", inputs={"X": reduced_sum}, outputs={"Out": rsquare})

    # TODO(caoying) the current elementwise_mul operator does not support a
    # general broadcast rule which broadcasts input(Y) to have the same
    # dimension with Input(X) starting from a specified dimension. So this
2496
    # exanpsion is requred. Once a general broadcast rule is spported, this
C
caoying03 已提交
2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513
    # expanding canbe removed.
    rsquare_expanded = helper.create_tmp_variable(dtype=x.dtype)
    expand_times = [1] * len(x.shape)
    expand_times[axis] = int(x.shape[axis])
    helper.append_op(
        type="expand",
        inputs={"X": rsquare},
        outputs={"Out": rsquare_expanded},
        attrs={"expand_times": expand_times})

    out = helper.create_tmp_variable(dtype=x.dtype)
    helper.append_op(
        type="elementwise_mul",
        inputs={"X": x,
                "Y": rsquare_expanded},
        outputs={"Out": out})
    return out
2514 2515


2516
def matmul(x, y, transpose_x=False, transpose_y=False, name=None):
G
guosheng 已提交
2517
    """
Y
ying 已提交
2518 2519 2520 2521
    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
2522

C
chengduoZH 已提交
2523
    The actual behavior depends on the shapes of :math:`x`, :math:`y` and the
2524
    flag values of :attr:`transpose_x`, :attr:`transpose_y`. Specifically:
G
guosheng 已提交
2525

2526 2527 2528 2529 2530
    - 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
2531
      :math:`[1, D]` in transposed form.
G
guosheng 已提交
2532

C
chengduoZH 已提交
2533
    - After transpose, the two tensors are 2-D or n-D and matrix multiplication
2534
      performs in the following way.
G
guosheng 已提交
2535

2536
      - If both are 2-D, they are multiplied like conventional matrices.
C
chengduoZH 已提交
2537
      - If either is n-D, it is treated as a stack of matrices residing in the
Y
ying 已提交
2538
        last two dimensions and a batched matrix multiply supporting broadcast
2539
        applies on the two tensors.
G
guosheng 已提交
2540

Y
ying 已提交
2541 2542
    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
2543
    removed after matrix multiplication.
G
guosheng 已提交
2544 2545 2546

    Args:
        x (Variable): The input variable which is a Tensor or LoDTensor.
2547 2548 2549
        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.
2550
        name(str|None): A name for this layer(optional). If set None, the layer
2551
            will be named automatically.
G
guosheng 已提交
2552 2553

    Returns:
2554
        Variable: The product Tensor variable.
G
guosheng 已提交
2555

G
guosheng 已提交
2556 2557 2558
    Examples:
        .. code-block:: python

2559
            # Examples to clarify shapes of the inputs and output
C
chengduoZH 已提交
2560 2561
            # x: [B, ..., M, K], y: [B, ..., K, N]
            fluid.layers.matmul(x, y)  # out: [B, ..., M, N]
Y
ying 已提交
2562

2563 2564
            # x: [B, M, K], y: [B, K, N]
            fluid.layers.matmul(x, y)  # out: [B, M, N]
Y
ying 已提交
2565

2566 2567
            # x: [B, M, K], y: [K, N]
            fluid.layers.matmul(x, y)  # out: [B, M, N]
Y
ying 已提交
2568

2569 2570
            # x: [M, K], y: [K, N]
            fluid.layers.matmul(x, y)  # out: [M, N]
Y
ying 已提交
2571 2572 2573 2574

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

2575 2576
            # x: [K], y: [K]
            fluid.layers.matmul(x, y)  # out: [1]
2577

Y
ying 已提交
2578
            # x: [M], y: [N]
2579
            fluid.layers.matmul(x, y, True, True)  # out: [M, N]
G
guosheng 已提交
2580
    """
Y
ying 已提交
2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592

    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
2593
            y_shape = y_shape + [1]
Y
ying 已提交
2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609

        # 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)

2610
    helper = LayerHelper('matmul', **locals())
Y
ying 已提交
2611
    out = helper.create_tmp_variable(dtype=x.dtype)
G
guosheng 已提交
2612
    helper.append_op(
2613 2614 2615 2616 2617 2618 2619
        type='matmul',
        inputs={'X': x,
                'Y': y},
        outputs={'Out': out},
        attrs={'transpose_X': transpose_x,
               'transpose_Y': transpose_y})
    return out
2620 2621


2622
def topk(input, k, name=None):
Q
qingqing01 已提交
2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637
    """
    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.
2638 2639
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
Q
qingqing01 已提交
2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670

    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
2671
def edit_distance(input, label, normalized=True, ignored_tokens=None,
W
wanghaoshuang 已提交
2672
                  name=None):
2673
    """
Y
ying 已提交
2674 2675 2676 2677 2678 2679 2680 2681 2682
    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
2683

Y
ying 已提交
2684
    "kitten" -> "sitten" -> "sittin" -> "sitting"
W
wanghaoshuang 已提交
2685

Y
ying 已提交
2686 2687 2688 2689
    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
2690

Y
ying 已提交
2691 2692 2693
    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
2694

2695 2696 2697 2698 2699
    Args:

        input(Variable): The indices for hypothesis strings.

        label(Variable): The indices for reference strings.
W
wanghaoshuang 已提交
2700

Y
ying 已提交
2701 2702
        normalized(bool): Indicated whether to normalize the edit distance by
                          the length of reference string.
2703

Y
ying 已提交
2704 2705
        ignored_tokens(list of int): Tokens that should be removed before
                                     calculating edit distance.
2706

W
wanghaoshuang 已提交
2707
    Returns:
W
wanghaoshuang 已提交
2708
        Variable: sequence-to-sequence edit distance in shape [batch_size, 1].
W
wanghaoshuang 已提交
2709 2710 2711 2712 2713

    Examples:
        .. code-block:: python

            x = fluid.layers.data(name='x', shape=[8], dtype='float32')
2714 2715
            y = fluid.layers.data(name='y', shape=[7], dtype='float32')

2716
            cost = fluid.layers.edit_distance(input=x,label=y)
2717
    """
2718
    helper = LayerHelper("edit_distance", **locals())
2719

2720
    # remove some tokens from input and labels
W
wanghaoshuang 已提交
2721
    if ignored_tokens is not None and len(ignored_tokens) > 0:
2722 2723 2724 2725 2726 2727 2728
        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
2729
            attrs={"tokens": ignored_tokens})
2730 2731 2732 2733 2734
        input = erased_input

        helper.append_op(
            type="sequence_erase",
            inputs={"X": [label]},
W
whs 已提交
2735
            outputs={"Out": [erased_label]},
W
wanghaoshuang 已提交
2736
            attrs={"tokens": ignored_tokens})
2737 2738
        label = erased_label

2739 2740
    # edit distance op
    edit_distance_out = helper.create_tmp_variable(dtype="int64")
2741
    sequence_num = helper.create_tmp_variable(dtype="int64")
2742 2743 2744 2745
    helper.append_op(
        type="edit_distance",
        inputs={"Hyps": [input],
                "Refs": [label]},
2746 2747
        outputs={"Out": [edit_distance_out],
                 "SequenceNum": [sequence_num]},
2748 2749
        attrs={"normalized": normalized})

2750
    return edit_distance_out, sequence_num
2751 2752 2753 2754 2755


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
2756 2757 2758 2759
    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.
2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788

    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
2789 2790 2791 2792 2793 2794
        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).
2795

Y
ying 已提交
2796 2797 2798
        blank(int): the blank label index of Connectionist Temporal
                    Classification (CTC) loss, which is in thehalf-opened
                    interval [0, num_classes + 1).
2799 2800

    Returns:
2801
        Variable: CTC greedy decode result. If all the sequences in result were
2802
        empty, the result LoDTensor will be [-1] with LoD [[0]] and dims [1, 1].
2803 2804 2805 2806 2807

    Examples:
        .. code-block:: python

            x = fluid.layers.data(name='x', shape=[8], dtype='float32')
W
wanghaoshuang 已提交
2808

2809
            cost = fluid.layers.ctc_greedy_decoder(input=x, blank=0)
W
wanghaoshuang 已提交
2810
    """
2811
    helper = LayerHelper("ctc_greedy_decoder", **locals())
Q
qingqing01 已提交
2812
    _, topk_indices = topk(input, k=1)
2813 2814 2815 2816 2817 2818

    # 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
2819
        outputs={"Output": [ctc_out]},
2820 2821
        attrs={"merge_repeated": True,
               "blank": blank})
2822
    return ctc_out
2823 2824


F
fengjiayi 已提交
2825
def warpctc(input, label, blank=0, norm_by_times=False):
W
wanghaoshuang 已提交
2826
    """
2827 2828
    An operator integrating the open source Warp-CTC library
    (https://github.com/baidu-research/warp-ctc)
W
wanghaoshuang 已提交
2829
    to compute Connectionist Temporal Classification (CTC) loss.
2830 2831
    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
2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844
    input tensor.

    Args:
       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.
2845
       blank: (int, default: 0), the blank label index of Connectionist
W
wanghaoshuang 已提交
2846 2847
         Temporal Classification (CTC) loss, which is in the
         half-opened interval [0, num_classes + 1).
2848
       norm_by_times: (bool, default: false), whether to normalize
W
wanghaoshuang 已提交
2849
       the gradients by the number of time-step, which is also the
2850 2851
       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
2852 2853

    Returns:
2854 2855
        Variable: The Connectionist Temporal Classification (CTC) loss,
        which is a 2-D Tensor of the shape [batch_size, 1].
W
wanghaoshuang 已提交
2856 2857 2858

    Examples:
        .. code-block:: python
2859 2860 2861 2862
            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
2863 2864 2865
            cost = layers.warpctc(input=y_predict, label=y)

    """
F
fengjiayi 已提交
2866
    helper = LayerHelper('warpctc', **locals())
W
wanghaoshuang 已提交
2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877
    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
2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931


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:
       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.

    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
2932 2933


2934
@autodoc()
Y
Yang Yu 已提交
2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960
def nce(input,
        label,
        num_total_classes,
        sample_weight=None,
        param_attr=None,
        bias_attr=None,
        num_neg_samples=None):
    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
2961 2962 2963 2964 2965 2966 2967 2968 2969
    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
2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985

    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
2986
    return cost / (num_neg_samples + 1)
2987 2988


Y
fix ci.  
ying 已提交
2989
def transpose(x, perm, name=None):
Y
ying 已提交
2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008
    """
    **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:
       input (Variable): (Tensor), A Tensor.
       perm (list): A permutation of the dimensions of `input`.

    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
3009
            x_transposed = layers.transpose(x, perm=[1, 0, 2])
Y
ying 已提交
3010 3011
    """

Y
fix ci.  
ying 已提交
3012
    if len(perm) != len(x.shape):
Y
ying 已提交
3013 3014 3015
        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
3016 3017 3018 3019 3020 3021
    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
3022 3023

    helper = LayerHelper('transpose', **locals())
Y
fix ci.  
ying 已提交
3024
    out = helper.create_tmp_variable(x.dtype)
Y
ying 已提交
3025 3026
    helper.append_op(
        type='transpose',
Y
fix ci.  
ying 已提交
3027
        inputs={'X': [x]},
Y
ying 已提交
3028 3029 3030
        outputs={'Out': [out]},
        attrs={'axis': perm})
    return out
3031 3032


3033
def im2sequence(input, filter_size=1, stride=1, padding=0, name=None):
3034
    """
3035 3036 3037 3038 3039 3040 3041
    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:
3042 3043 3044 3045 3046 3047 3048 3049 3050 3051

    .. 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
3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069

        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.

3070 3071 3072
        name (int): The name of this layer. It is optional.

    Returns:
W
wanghaoshuang 已提交
3073 3074 3075 3076 3077
        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.
3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106

    Examples:

    As an example:

        .. 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
3107 3108 3109
            filter = [2, 2]
            stride = [1, 1]
            padding = [0, 0]
3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129

            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]]

        The simple usage is:

        .. code-block:: python

3130 3131
            output = fluid.layers.im2sequence(
                input=layer, stride=[1, 1], filter_size=[2, 2])
3132 3133

    """
W
wanghaoshuang 已提交
3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144

    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])

3145
    helper = LayerHelper('im2sequence', **locals())
3146 3147
    out = helper.create_tmp_variable(dtype=helper.input_dtype())
    helper.append_op(
3148
        type='im2sequence',
3149 3150 3151
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
W
wanghaoshuang 已提交
3152 3153 3154
            'kernels': filter_size,
            'strides': stride,
            'paddings': padding,
3155 3156
        })
    return out
3157 3158


3159 3160 3161 3162
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
3163
    equation of row convolution is as follows:
3164 3165 3166 3167 3168 3169 3170

    .. 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
3171
    * :math:`\\tau`: Future context size.
3172 3173 3174 3175 3176 3177 3178 3179 3180 3181
    * :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
3182 3183
        future_context_size (int): Future context size. Please note, the shape
            of convolution kernel is [future_context_size + 1, D].
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
        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
3209
    return helper.append_activation(out)
3210 3211


3212 3213 3214 3215
def multiplex(inputs, index):
    """
    **Multiplex Layer**

Y
yangyaming 已提交
3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230
    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]`.
3231 3232

    Args:
Y
yangyaming 已提交
3233 3234
       inputs (list): A list of variables to gather from. All variables have the
                same shape and the rank is at least 2.
3235
       index (Variable): Tensor<int32>, index variable which is a 2-D tensor
Y
yangyaming 已提交
3236
                with shape [M, 1] where M is the batch size.
3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249

    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
3250 3251 3252 3253 3254 3255

    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)
3256 3257 3258 3259 3260 3261
    helper.append_op(
        type='multiplex',
        inputs={'X': inputs,
                'Ids': index},
        outputs={'Out': [out]})
    return out
3262 3263 3264 3265 3266


def softmax_with_cross_entropy(logits, label, soft_label=False):
    """
    **Softmax With Cross Entropy Operator.**
3267

3268 3269 3270 3271
    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.
3272

3273 3274 3275
    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.
3276

3277 3278 3279
    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.
3280

3281
    The equation is as follows:
3282

3283
    1) Hard label (one-hot label, so every sample has exactly one class)
3284

3285 3286 3287 3288
    .. math::

        loss_j =  -\\text{logit}_{label_j} +
        \\log\\left(\\sum_{i=0}^{K}\\exp(\\text{logit}_i)\\right), j = 1,..., K
3289

3290 3291 3292
    2) Soft label (each sample can have a distribution over all classes)

    .. math::
3293

3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314
        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
3315 3316
            out = fluid.layers.softmax_with_cross_entropy(
                logits=fc, label=label)
3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334
    """
    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
3335
    This operator computes the smooth L1 loss for X and Y.
3336
    The operator takes the first dimension of X and Y as batch size.
Q
qingqing01 已提交
3337
    For each instance, it computes the smooth L1 loss element by element first
3338
    and then sums all the losses. So the shape of Out is [batch_size, 1].
3339

3340 3341
    Args:
        x (Variable): A tensor with rank at least 2. The input value of smooth
Q
qingqing01 已提交
3342
            L1 loss op with shape [batch_size, dim1, ..., dimN].
3343
        y (Variable): A tensor with rank at least 2. The target value of smooth
Q
qingqing01 已提交
3344
            L1 loss op with same shape as x.
3345 3346 3347 3348 3349 3350
        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
3351
            the out smooth L1 loss will be multiplied by this tensor element
3352
            by element.
Q
qingqing01 已提交
3353
        sigma (float|None): Hyper parameter of smooth L1 loss op. A float scalar
3354 3355
            with default value 1.0.
    Returns:
Q
qingqing01 已提交
3356
        Variable: A tensor with rank be 2. The output smooth L1 loss with
3357 3358 3359 3360 3361 3362
            shape [batch_size, 1].

    Examples:
        .. code-block:: python

            data = fluid.layers.data(name='data', shape=[128], dtype='float32')
F
stash  
fengjiayi 已提交
3363 3364
            label = fluid.layers.data(
                name='label', shape=[100], dtype='float32')
3365
            fc = fluid.layers.fc(input=data, size=100)
F
fengjiayi 已提交
3366
            out = fluid.layers.smooth_l1(x=fc, y=label)
3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382
    """
    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
3383 3384 3385 3386 3387 3388 3389 3390 3391


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
3392
        input(variable):  A Tensor/LodTensor of indices, last dimension must be 1.
3393 3394 3395 3396 3397 3398
        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
3399 3400
        .. code-block:: python

3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421
        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
3422 3423


Y
Yu Yang 已提交
3424
def autoincreased_step_counter(counter_name=None, begin=1, step=1):
Y
Yu Yang 已提交
3425
    """
Y
Yu Yang 已提交
3426
    NOTE: The counter will be automatically increased by 1 every mini-batch
Y
Yu Yang 已提交
3427
    Return the run counter of the main program, which is started with 1.
Y
Yu Yang 已提交
3428 3429 3430 3431 3432 3433

    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.

Y
Yu Yang 已提交
3434 3435 3436
    Returns(Variable): The global run counter.
    """
    helper = LayerHelper('global_step_counter')
Y
Yu Yang 已提交
3437 3438
    if counter_name is None:
        counter_name = '@STEP_COUNTER@'
Y
Yu Yang 已提交
3439 3440 3441 3442 3443
    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
3444
                value=begin - 1, force_cpu=True))
Y
Yu Yang 已提交
3445 3446 3447
        helper.main_program.global_block().prepend_op(
            type='increment',
            inputs={'X': [counter]},
Y
Yu Yang 已提交
3448 3449
            outputs={'Out': [counter]},
            attrs={'step': float(step)})
Y
Yu Yang 已提交
3450 3451 3452
        counter.stop_gradient = True

    return counter
Y
yangyaming 已提交
3453 3454


3455
def reshape(x, shape, actual_shape=None, act=None, inplace=True, name=None):
C
caoying03 已提交
3456
    """
C
caoying03 已提交
3457 3458
    Gives a new shape to the input Tensor without changing its data.

3459 3460 3461 3462 3463
    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
3464

3465
    Some tricks exist when specifying the target shape.
C
caoying03 已提交
3466

3467 3468 3469 3470
    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.

3471
    2. 0 means the actual dimension value is going to be copied from the
3472 3473 3474 3475
    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
3476 3477

    1. Given a 3-D tensor x with a shape [2, 4, 6], and the target shape
W
wanghaoshuang 已提交
3478
    is [6, 8], the reshape operator will transform x into a 2-D tensor with
3479
    shape [6, 8] and leaving x's data unchanged.
C
caoying03 已提交
3480

3481
    2. Given a 3-D tensor x with a shape [2, 4, 6], and the target shape
3482 3483
    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
3484 3485
    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
3486
    dimensions.
C
caoying03 已提交
3487

3488
    3. Given a 3-D tensor x with a shape [2, 4, 6], and the target shape
3489 3490 3491 3492
    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
3493 3494 3495 3496 3497

    Args:
        input(variable): The input tensor.
        shape(list): The new shape. At most one dimension of the new shape can
                     be -1.
3498 3499 3500 3501 3502
        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
3503 3504 3505 3506 3507 3508 3509 3510 3511
        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.

    Returns(variable): The output tensor.

    Examples:
        .. code-block:: python
G
guosheng 已提交
3512

3513
            data = fluid.layers.data(
3514
                name='data', shape=[2, 4, 6], dtype='float32')
C
caoying03 已提交
3515
            reshaped = fluid.layers.reshape(
3516
                x=data, shape=[-1, 0, 3, 2], act='tanh', inplace=True)
C
caoying03 已提交
3517 3518 3519 3520 3521
    """

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

3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536
    # 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
3537 3538 3539 3540
    helper = LayerHelper("reshape", **locals())
    reshaped = helper.create_tmp_variable(dtype=x.dtype)
    helper.append_op(
        type="reshape",
3541 3542 3543
        inputs={"X": x,
                "Shape": actual_shape}
        if isinstance(actual_shape, Variable) else {"X": x},
C
caoying03 已提交
3544 3545 3546 3547 3548
        attrs={"shape": shape,
               "inplace": inplace},
        outputs={"Out": reshaped})

    return helper.append_activation(reshaped)
3549 3550


Y
yangyaming 已提交
3551
def lod_reset(x, y=None, target_lod=None):
Y
yangyaming 已提交
3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643
    """
    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
3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685


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::

        Output(i, x, y) = Input(i, x, y) / \left(
        k + \alpha \sum\limits^{\min(C, c + n/2)}_{j = \max(0, c - n/2)}
        (Input(j, x, y))^2 \right)^{\beta}

    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
3686 3687
          data = fluid.layers.data(
              name="data", shape=[3, 112, 112], dtype="float32")
D
dragonwarrior 已提交
3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714
          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
3715 3716 3717 3718


def pad(x, paddings, pad_value=0., name=None):
    """
G
guosheng 已提交
3719
    Pads a tensor with a constant value given by :attr:`pad_value`, and the
W
wanghaoshuang 已提交
3720
    padded width is specified by :attr:`paddings`.
G
guosheng 已提交
3721

G
guosheng 已提交
3722 3723 3724 3725
    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
3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747

    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
3748
                         The length of :attr:paddings must be
G
guosheng 已提交
3749 3750 3751 3752 3753 3754 3755 3756 3757 3758
                         :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
3759

G
guosheng 已提交
3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773
            # 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
3774 3775 3776 3777 3778 3779 3780 3781 3782


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
3783 3784
    called label-smoothing regularization (LSR).

3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807
    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
3808
                              be :math:`(1, class\_num)`.
3809 3810
        epsilon(float): The weight used to mix up the original ground-truth
                        distribution and the fixed distribution.
3811
        dtype(np.dtype|core.VarDesc.VarType|str): The type of data : float32,
3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838
                                                  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
3839 3840 3841 3842


def roi_pool(input, rois, pooled_height=1, pooled_width=1, spatial_scale=1.0):
    """
3843
    Region of interest pooling (also known as RoI pooling) is to perform
3844 3845
        is to perform max pooling on inputs of nonuniform sizes to obtain
        fixed-size feature maps (e.g. 7*7).
3846 3847 3848 3849
    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
3850 3851 3852 3853 3854 3855 3856
        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)
3857 3858
                         is the top left coordinates, and (x2, y2) is the
                         bottom right coordinates. The num_rois is the
3859 3860 3861 3862 3863 3864 3865 3866
                         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:
3867
        pool_out (Variable): The output is a 4-D tensor of the shape
3868 3869 3870
                             (num_rois, channels, pooled_h, pooled_w).

    Examples:
3871 3872
        .. code-block:: python

3873
            pool_out = fluid.layers.roi_pool(input=x, rois=rois, 7, 7, 1.0)
3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890
    """
    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
3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919


def dice_loss(input, label, epsilon=0.00001):
    """
    **Dice loss Layer**
    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:
3920 3921
        .. code-block:: python

W
whs 已提交
3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932
            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)
3933 3934


3935
def upsampling_bilinear2d(input, out_shape=None, scale=None, name=None):
3936
    """
3937 3938
    The mathematical meaning of upsampling_bilinear2d is also called
    Bilinear interpolation.
3939 3940 3941
    Bilinear interpolation is an extension of linear interpolation for
    interpolating functions of two variables (e.g. H-direction and
    W-direction in this layer) on a rectilinear 2D grid.
F
stash  
fengjiayi 已提交
3942

3943 3944
    For details, please refer to Wikipedia:
    https://en.wikipedia.org/wiki/Bilinear_interpolation
F
stash  
fengjiayi 已提交
3945

3946 3947 3948 3949
    Args:
        input (Variable): The input tensor of bilinear interpolation,
                          This is a 4-D tensor of the shape
                          (num_batches, channels, in_h, in_w).
3950 3951 3952 3953 3954 3955 3956
        out_shape(list|tuple|None): Output shape of bilinear interpolation
                                    layer, the shape is (out_h, out_w).
                                    Default: None
        scale(int|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
3957 3958 3959 3960 3961 3962
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.

    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
3963

3964 3965 3966
    Examples:
        .. code-block:: python

3967
            out = fluid.layers.bilinear_interp(input, out_shape=[12, 12])
3968
    """
3969 3970
    if out_shape is None and scale is None:
        raise ValueError("One of out_shape and scale must not be None")
3971 3972
    helper = LayerHelper('bilinear_interp', **locals())
    dtype = helper.input_dtype()
3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987

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

    if out_shape is not None:
        if not (_is_list_or_turple_(out_shape) and len(out_shape) == 2):
            raise ValueError('out_shape should be a list or tuple ',
                             'with length 2, (out_h, out_w).')
        out_shape = list(map(int, out_shape))
        out_h = out_shape[0]
        out_w = out_shape[1]
    else:
        out_h = int(input.shape[2] * scale)
        out_w = int(input.shape[3] * scale)

3988 3989 3990 3991 3992 3993 3994 3995
    out = helper.create_tmp_variable(dtype)
    helper.append_op(
        type="bilinear_interp",
        inputs={"X": input},
        outputs={"Out": out},
        attrs={"out_h": out_h,
               "out_w": out_w})
    return out
F
stash  
fengjiayi 已提交
3996 3997


F
fengjiayi 已提交
3998
def random_crop(input, shape, seed=1):
F
stash  
fengjiayi 已提交
3999 4000 4001 4002
    helper = LayerHelper("random_crop", **locals())
    dtype = helper.input_dtype()
    out = helper.create_tmp_variable(dtype)
    if isinstance(seed, int):
F
fengjiayi 已提交
4003
        seed_value = seed
F
fengjiayi 已提交
4004 4005 4006 4007 4008 4009 4010 4011 4012 4013
        seed = helper.create_tmp_variable(dtype="int64")
        helper.append_op(
            type="fill_constant",
            inputs={},
            outputs={"Out": seed},
            attrs={
                "dtype": seed.dtype,
                "shape": [1],
                "value": float(seed_value)
            })
F
stash  
fengjiayi 已提交
4014 4015
    elif not isinstance(seed, Variable):
        raise ValueError("'seed' must be a Variable or an int.")
F
fengjiayi 已提交
4016
    seed_out = helper.create_tmp_variable(dtype="int64")
F
stash  
fengjiayi 已提交
4017 4018 4019 4020 4021 4022 4023 4024
    helper.append_op(
        type="random_crop",
        inputs={"X": input,
                "Seed": seed},
        outputs={"Out": out,
                 "SeedOut": seed_out},
        attrs={"shape": shape})
    return out