nn.py 131.1 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 63 64
    'l2_normalize',
    'matmul',
    'warpctc',
    'sequence_reshape',
65
    'transpose',
66
    'im2sequence',
67
    'nce',
Q
Qiao Longfei 已提交
68
    'beam_search',
69
    'row_conv',
70
    'multiplex',
G
guosheng 已提交
71
    'layer_norm',
72 73
    'softmax_with_cross_entropy',
    'smooth_l1',
74
    'one_hot',
Y
Yu Yang 已提交
75
    'autoincreased_step_counter',
C
caoying03 已提交
76
    'reshape',
Y
yangyaming 已提交
77
    'lod_reset',
D
dragonwarrior 已提交
78
    'lrn',
Y
Yu Yang 已提交
79 80 81 82 83 84 85 86
]


def fc(input,
       size,
       num_flatten_dims=1,
       param_attr=None,
       bias_attr=None,
87
       use_mkldnn=False,
Y
Yu Yang 已提交
88
       act=None,
89
       name=None):
Y
Yu Yang 已提交
90
    """
91
    **Fully Connected Layer**
Y
Yu Yang 已提交
92

C
caoying03 已提交
93
    The fully connected layer can take multiple tensors as its inputs. It
R
ranqiu 已提交
94 95 96 97 98 99
    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
100
    if activation is not None, it will be applied to the output as well.
C
caoying03 已提交
101

C
caoying03 已提交
102
    This process can be formulated as follows:
103 104 105

    .. math::

106
        Out = Act({\sum_{i=0}^{N-1}X_iW_i + b})
107 108 109

    In the above equation:

C
caoying03 已提交
110 111 112 113
    * :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).
114
    * :math:`Act`: The activation function.
C
caoying03 已提交
115
    * :math:`Out`: The output tensor.
Y
Yu Yang 已提交
116 117

    Args:
R
ranqiu 已提交
118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135
        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.
        name (str, default None): The name of this layer.
Y
Yu Yang 已提交
136

137
    Returns:
R
ranqiu 已提交
138
        A tensor variable storing the transformation result.
139 140

    Raises:
C
caoying03 已提交
141
        ValueError: If rank of the input tensor is less than 2.
142 143 144 145

    Examples:
        .. code-block:: python

C
caoying03 已提交
146
          data = fluid.layers.data(name="data", shape=[32, 32], dtype="float32")
147
          fc = fluid.layers.fc(input=data, size=1000, act="tanh")
Y
Yu Yang 已提交
148
    """
C
caoying03 已提交
149

C
caoying03 已提交
150
    helper = LayerHelper("fc", **locals())
Y
Yu Yang 已提交
151 152 153 154 155 156 157 158 159

    dtype = helper.input_dtype()

    mul_results = []
    for input_var, param_attr in helper.iter_inputs_and_params():
        input_shape = input_var.shape
        param_shape = [
            reduce(lambda a, b: a * b, input_shape[num_flatten_dims:], 1)
        ] + [size]
Y
ying 已提交
160

Y
Yu Yang 已提交
161 162 163 164 165
        w = helper.create_parameter(
            attr=param_attr, shape=param_shape, dtype=dtype, is_bias=False)
        tmp = helper.create_tmp_variable(dtype)
        helper.append_op(
            type="mul",
Q
Qiao Longfei 已提交
166 167
            inputs={"X": input_var,
                    "Y": w},
Y
Yu Yang 已提交
168
            outputs={"Out": tmp},
169 170 171 172 173
            attrs={
                "x_num_col_dims": num_flatten_dims,
                "y_num_col_dims": 1,
                'use_mkldnn': use_mkldnn
            })
Y
Yu Yang 已提交
174 175 176 177 178 179 180 181 182 183
        mul_results.append(tmp)

    # sum
    if len(mul_results) == 1:
        pre_bias = mul_results[0]
    else:
        pre_bias = helper.create_tmp_variable(dtype)
        helper.append_op(
            type="sum", inputs={"X": mul_results}, outputs={"Out": pre_bias})
    # add bias
184
    pre_activation = helper.append_bias_op(pre_bias, dim_start=num_flatten_dims)
Y
Yu Yang 已提交
185 186 187 188
    # add activation
    return helper.append_activation(pre_activation)


189 190 191 192 193 194
def embedding(input,
              size,
              is_sparse=False,
              padding_idx=None,
              param_attr=None,
              dtype='float32'):
Y
Yu Yang 已提交
195
    """
196 197
    **Embedding Layer**

198
    This layer is used to lookup embeddings of IDs, provided by :attr:`input`, in
199 200
    a lookup table. The result of this lookup is the embedding of each ID in the
    :attr:`input`.
201 202 203

    All the input variables are passed in as local variables to the LayerHelper
    constructor.
Y
Yu Yang 已提交
204 205

    Args:
206 207 208 209 210 211 212
        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
213 214
            with zeros whenever lookup encounters it in :attr:`input`. If
            :math:`padding_idx < 0`, the padding_idx to use in lookup is
215 216
            :math:`size[0] + dim`.
        param_attr(ParamAttr): Parameters for this layer
217
        dtype(np.dtype|core.VarDesc.VarType|str): The type of data : float32, float_16, int etc
Y
Yu Yang 已提交
218

219 220 221
    Returns:
        Variable: The tensor variable storing the embeddings of the \
                  supplied inputs.
Y
Yu Yang 已提交
222

223 224
    Examples:
        .. code-block:: python
Y
Yu Yang 已提交
225

C
chengduoZH 已提交
226
          dict_size = len(dataset.ids)
227
          data = fluid.layers.data(name='ids', shape=[32, 32], dtype='float32')
C
chengduoZH 已提交
228
          fc = fluid.layers.embedding(input=data, size=[dict_size, 16])
Y
Yu Yang 已提交
229 230 231 232 233 234
    """

    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)
235 236
    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
237 238 239 240 241
    helper.append_op(
        type='lookup_table',
        inputs={'Ids': input,
                'W': w},
        outputs={'Out': tmp},
242 243
        attrs={'is_sparse': is_sparse,
               'padding_idx': padding_idx})
Y
Yu Yang 已提交
244 245 246 247 248 249 250 251 252 253 254 255 256
    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',
257 258
                 dtype='float32',
                 name=None):
Y
Yibing Liu 已提交
259 260 261 262 263 264
    """
    **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
265
    .. math::
Y
Yibing Liu 已提交
266

267
        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
268

269
        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
270

271
        \\tilde{c_t} & = act_g(W_{cx}x_t + W_{ch}h_{t-1} + b_c)
Y
Yibing Liu 已提交
272

273 274 275
        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
276

Y
Yibing Liu 已提交
277
        h_t & = o_t \odot act_h(c_t)
Y
Yibing Liu 已提交
278

279
    where the :math:`W` terms denote weight matrices (e.g. :math:`W_{xi}` is
280
    the matrix of weights from the input gate to the input), :math:`W_{ic}, \
281 282 283
    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
284
    gate bias vector), :math:`\sigma` is the non-linear activations, such as
285 286
    logistic sigmoid function, and :math:`i, f, o` and :math:`c` are the input
    gate, forget gate, output gate, and cell activation vectors, respectively,
287 288
    all of which have the same size as the cell output activation vector :math:`h`.

289 290 291 292
    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
293 294 295
    the previous hidden state.

    Set `use_peepholes` to `False` to disable peephole connection. The formula
Y
Yibing Liu 已提交
296 297 298
    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
299 300 301
    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
302 303

    Args:
304 305 306 307
        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
308 309
                         mini-batch, D is the hidden size.
        size(int): 4 * hidden size.
310
        param_attr(ParamAttr|None): The parameter attribute for the learnable
311
                               hidden-hidden weights.
Y
Yibing Liu 已提交
312 313 314

                               - Weights = {:math:`W_{ch}, W_{ih}, \
                                                W_{fh}, W_{oh}`}
315 316 317
                               - The shape is (D x 4D), where D is the hidden
                                 size.
        bias_attr(ParamAttr|None): The bias attribute for the learnable bias
318 319 320
                              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
321

322
                              1. `use_peepholes = False`
Y
Yibing Liu 已提交
323
                                - Biases = {:math:`b_c, b_i, b_f, b_o`}.
324
                                - The shape is (1 x 4D).
325
                              2. `use_peepholes = True`
Y
Yibing Liu 已提交
326 327
                                - Biases = { :math:`b_c, b_i, b_f, b_o, W_{ic}, \
                                                 W_{fc}, W_{oc}`}.
328
                                - The shape is (1 x 7D).
329
        use_peepholes(bool): Whether to enable diagonal/peephole connections,
Y
Yibing Liu 已提交
330 331
                             default `True`.
        is_reverse(bool): Whether to compute reversed LSTM, default `False`.
332 333
        gate_activation(str): The activation for input gate, forget gate and
                              output gate. Choices = ["sigmoid", "tanh", "relu",
Y
Yibing Liu 已提交
334
                              "identity"], default "sigmoid".
335
        cell_activation(str): The activation for cell output. Choices = ["sigmoid",
Y
Yibing Liu 已提交
336 337 338 339 340
                              "tanh", "relu", "identity"], default "tanh".
        candidate_activation(str): The activation for candidate hidden state.
                              Choices = ["sigmoid", "tanh", "relu", "identity"],
                              default "tanh".
        dtype(str): Data type. Choices = ["float32", "float64"], default "float32".
341 342
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.
Y
Yibing Liu 已提交
343 344

    Returns:
Y
Yibing Liu 已提交
345 346
        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
347

Y
Yibing Liu 已提交
348
    Examples:
Y
Yibing Liu 已提交
349 350
        .. code-block:: python

Y
Yibing Liu 已提交
351 352
            hidden_dim = 512
            forward_proj = fluid.layers.fc(input=input_seq, size=hidden_dim * 4,
353
                                           act=None, bias_attr=None)
Y
Yibing Liu 已提交
354 355
            forward, _ = fluid.layers.dynamic_lstm(
                input=forward_proj, size=hidden_dim * 4, use_peepholes=False)
Y
Yibing Liu 已提交
356
    """
357

Y
Yu Yang 已提交
358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393
    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
394 395 396 397 398 399 400 401 402 403 404
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',
405 406
                  dtype='float32',
                  name=None):
Y
Yibing Liu 已提交
407 408 409
    """
    **Dynamic LSTMP Layer**

410 411 412 413 414 415
    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
416 417 418 419 420

    The formula is as follows:

    .. math::

421
        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
422

423
        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
424

425
        \\tilde{c_t} & = act_g(W_{cx}x_t + W_{cr}r_{t-1} + b_c)
Y
Yibing Liu 已提交
426

427
        o_t & = \sigma(W_{ox}x_{t} + W_{or}r_{t-1} + W_{oc}c_t + b_o)
Y
Yibing Liu 已提交
428

429
        c_t & = f_t \odot c_{t-1} + i_t \odot \\tilde{c_t}
Y
Yibing Liu 已提交
430

431
        h_t & = o_t \odot act_h(c_t)
Y
Yibing Liu 已提交
432

433
        r_t & = \overline{act_h}(W_{rh}h_t)
Y
Yibing Liu 已提交
434

Y
Yibing Liu 已提交
435 436 437 438 439 440
    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, \
441
          we use vectors to reprenset these diagonal weight matrices.
Y
Yibing Liu 已提交
442
    * :math:`b`: Denotes bias vectors (e.g. :math:`b_i` is the input gate \
443
          bias vector).
Y
Yibing Liu 已提交
444 445 446
    * :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 \
447
          the same size as the cell output activation vector :math:`h`.
Y
Yibing Liu 已提交
448
    * :math:`h`: The hidden state.
449
    * :math:`r`: The recurrent projection of the hidden state.
Y
Yibing Liu 已提交
450 451
    * :math:`\\tilde{c_t}`: The candidate hidden state, whose \
          computation is based on the current input and previous hidden state.
452
    * :math:`\odot`: The element-wise product of the vectors.
Y
Yibing Liu 已提交
453
    * :math:`act_g` and :math:`act_h`: The cell input and cell output \
454
          activation functions and `tanh` is usually used for them.
Y
Yibing Liu 已提交
455 456
    * :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
457 458 459 460

    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.
461

Y
Yibing Liu 已提交
462 463 464 465 466 467 468 469 470 471 472 473
    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.
474
        param_attr(ParamAttr|None): The parameter attribute for the learnable
Y
Yibing Liu 已提交
475 476
                               hidden-hidden weight and projection weight.

477 478
                               - Hidden-hidden weight = {:math:`W_{ch}, W_{ih}, \
                                                W_{fh}, W_{oh}`}.
479 480
                               - 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
481 482
                                 size.
                               - Projection weight = {:math:`W_{rh}`}.
483 484
                               - 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
485 486 487 488 489 490
                              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`}.
491
                                - The shape is (1 x 4D).
Y
Yibing Liu 已提交
492 493 494
                              2. `use_peepholes = True`
                                - Biases = { :math:`b_c, b_i, b_f, b_o, W_{ic}, \
                                                 W_{fc}, W_{oc}`}.
495
                                - The shape is (1 x 7D).
Y
Yibing Liu 已提交
496 497 498 499 500 501 502 503 504 505 506 507 508 509 510
        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.
                              Choices = ["sigmoid", "tanh", "relu", "identity"],
                              default "tanh".
        proj_activation(str): The activation for projection output.
                              Choices = ["sigmoid", "tanh", "relu", "identity"],
                              default "tanh".
        dtype(str): Data type. Choices = ["float32", "float64"], default "float32".
511 512
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.
Y
Yibing Liu 已提交
513 514

    Returns:
515 516
        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
517 518 519 520 521
               (T x D), and both LoD is the same with the `input`.

    Examples:
        .. code-block:: python

Y
Yibing Liu 已提交
522
            hidden_dim, proj_dim = 512, 256
Y
Yibing Liu 已提交
523 524
            fc_out = fluid.layers.fc(input=input_seq, size=hidden_dim * 4,
                                     act=None, bias_attr=None)
525 526 527
            proj_out, _ = fluid.layers.dynamic_lstmp(input=fc_out,
                                                     size=hidden_dim * 4,
                                                     proj_size=proj_dim,
Y
Yibing Liu 已提交
528 529 530 531
                                                     use_peepholes=False,
                                                     is_reverse=True,
                                                     cell_activation="tanh",
                                                     proj_activation="tanh")
Y
Yibing Liu 已提交
532
    """
533

Y
Yibing Liu 已提交
534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579
    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
580 581 582 583 584 585 586 587 588 589 590
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**

591
    Refer to `Empirical Evaluation of Gated Recurrent Neural Networks on
G
guosheng 已提交
592
    Sequence Modeling <https://arxiv.org/abs/1412.3555>`_
593

G
guosheng 已提交
594 595 596 597 598 599 600 601 602
    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)
603

G
guosheng 已提交
604
        h_t & = (1-u_t) \odot h_{t-1} + u_t \odot \\tilde{h_t}
605

G
guosheng 已提交
606
    The :math:`\odot` is the element-wise product of the vectors. :math:`act_g`
607 608
    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
609 610 611 612
    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
613
    to use fully-connect layer before GRU layer.
G
guosheng 已提交
614 615

    Args:
616 617
        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
618
            Variable is a matrix with shape :math:`(T \\times 3D)`, where
619
            :math:`T` is the total time steps in this mini-batch, :math:`D`
G
guosheng 已提交
620 621
            is the hidden size.
        size(int): The dimension of the gru cell.
622
        param_attr(ParamAttr|None): The parameter attribute for the learnable
G
guosheng 已提交
623 624
            hidden-hidden weight matrix. Note:

625
            - The shape of the weight matrix is :math:`(T \\times 3D)`, where
G
guosheng 已提交
626
              :math:`D` is the hidden size.
627
            - All elements in the weight matrix can be divided into two parts.
G
guosheng 已提交
628
              The first part are weights of the update gate and reset gate with
629
              shape :math:`(D \\times 2D)`, and the second part are weights for
G
guosheng 已提交
630
              candidate hidden state with shape :math:`(D \\times D)`.
631
        bias_attr(ParamAttr): The parameter attribute for learnable the
G
guosheng 已提交
632
            hidden-hidden bias.
633
        is_reverse(bool): Whether to compute reversed GRU, default
G
guosheng 已提交
634 635 636
            :attr:`False`.
        gate_activation(str): The activation for update gate and reset gate.
            Choices = ["sigmoid", "tanh", "relu", "identity"], default "sigmoid".
637
        activation(str): The activation for candidate hidden state.
G
guosheng 已提交
638 639 640
            Choices = ["sigmoid", "tanh", "relu", "identity"], default "tanh".

    Returns:
G
guosheng 已提交
641 642
        Variable: The hidden state of GRU. The shape is :math:`(T \\times D)`, \
            and lod is the same with the input.
643

G
guosheng 已提交
644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686
    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
687 688 689 690 691 692
def gru_unit(input,
             hidden,
             size,
             weight=None,
             bias=None,
             activation='tanh',
693
             gate_activation='sigmoid'):
Y
Yu Yang 已提交
694
    """
695
    GRU unit layer. The equation of a gru step is:
Y
Yu Yang 已提交
696

697 698
        .. math::
            u_t & = actGate(xu_{t} + W_u h_{t-1} + b_u)
Y
Yu Yang 已提交
699

700
            r_t & = actGate(xr_{t} + W_r h_{t-1} + b_r)
Y
Yu Yang 已提交
701

702
            m_t & = actNode(xm_t + W_c dot(r_t, h_{t-1}) + b_m)
703

704
            h_t & = dot((1-u_t), m_t) + dot(u_t, h_{t-1})
705 706

    The inputs of gru unit includes :math:`z_t`, :math:`h_{t-1}`. In terms
707 708 709
    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
710 711
    connected layer has to be applied, such that :math:`z_t = W_{fc}x_t`.

712 713
    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
714 715 716
    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`.
717 718 719 720 721 722 723

    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.
        weight (ParamAttr): The weight parameters for gru unit. Default: None
        bias (ParamAttr): The bias parameters for gru unit. Default: None
724 725 726 727
        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
728

729 730 731 732 733 734
    Returns:
        tuple: The hidden value, reset-hidden value and gate values.

    Examples:

        .. code-block:: python
Y
Yu Yang 已提交
735

736
             # assuming we have x_t_data and prev_hidden of size=10
737
             x_t = fluid.layers.fc(input=x_t_data, size=30)
738 739
             hidden_val, r_h_val, gate_val = fluid.layers.gru_unit(input=x_t,
                                                    hidden = prev_hidden)
Y
Yu Yang 已提交
740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759

    """
    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
    if weight is None:
        weight = helper.create_parameter(
            attr=helper.param_attr, shape=[size, 3 * size], dtype=dtype)

    # create bias
Y
Yibing Liu 已提交
760

Y
Yu Yang 已提交
761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787
    if bias is None:
        bias_size = [1, 3 * size]
        bias = helper.create_parameter(
            attr=helper.bias_attr, shape=bias_size, dtype=dtype, is_bias=True)

    gate = helper.create_tmp_variable(dtype)
    reset_hidden_pre = helper.create_tmp_variable(dtype)
    updated_hidden = helper.create_tmp_variable(dtype)

    helper.append_op(
        type='gru_unit',
        inputs={'Input': input,
                'HiddenPrev': hidden,
                'Weight': weight},
        outputs={
            'Gate': gate,
            'ResetHiddenPrev': reset_hidden_pre,
            'Hidden': updated_hidden,
        },
        attrs={
            'activation': 0,
            'gate_activation': 1,
        })

    return updated_hidden, reset_hidden_pre, gate


788
def linear_chain_crf(input, label, param_attr=None):
Y
Yu Yang 已提交
789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813
    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


814
def crf_decoding(input, param_attr, label=None):
Y
Yu Yang 已提交
815 816 817 818 819 820 821 822 823 824 825 826 827
    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
828
def cos_sim(X, Y):
Y
Yu Yang 已提交
829 830 831 832
    """
    This function performs the cosine similarity between two tensors
    X and Y and returns that as the output.
    """
F
fengjiayi 已提交
833
    helper = LayerHelper('cos_sim', **locals())
Y
Yu Yang 已提交
834 835 836 837 838 839 840 841 842 843 844 845 846
    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


F
fengjiayi 已提交
847
def dropout(x, dropout_prob, is_test=False, seed=None):
848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875
    """
    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.

    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
876
    helper = LayerHelper('dropout', **locals())
877 878 879 880 881 882 883
    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]},
884 885 886 887 888 889
        attrs={
            'dropout_prob': dropout_prob,
            'is_test': is_test,
            'fix_seed': seed is not None,
            'seed': seed if seed is not None else 0
        })
890 891 892
    return out


F
fengjiayi 已提交
893
def cross_entropy(input, label, soft_label=False):
Y
Yu Yang 已提交
894
    """
Y
Yibing Liu 已提交
895 896
    **Cross Entropy Layer**

897 898 899
    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
900 901

    1) One-hot cross-entropy:
F
fengjiayi 已提交
902
        `soft_label = False`, `Label[i, 0]` indicates the class index for sample i:
Y
yangyaming 已提交
903

Y
Yibing Liu 已提交
904
        .. math::
Y
yangyaming 已提交
905

Y
Yibing Liu 已提交
906 907 908
            Y[i] = -\log(X[i, Label[i]])

    2) Soft-label cross-entropy:
F
fengjiayi 已提交
909 910
        `soft_label = True`, `Label[i, j]` indicates the soft label of class j
        for sample i:
Y
Yibing Liu 已提交
911 912 913 914 915

        .. math::

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

Y
Yibing Liu 已提交
916
       Please make sure that in this case the summation of each row of `label`
Y
Yibing Liu 已提交
917 918 919
       equals one.

    3) One-hot cross-entropy with vecterized `label`:
F
fengjiayi 已提交
920 921
         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
922
         to a one-hot cross-entropy with one-hot label representation.
Y
yangyaming 已提交
923

Y
Yibing Liu 已提交
924
    Args:
Y
yangyaming 已提交
925
        input (Variable|list):  a 2-D tensor with shape [N x D], where N is the
926 927 928 929
                                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
930
        label (Variable|list): the ground truth which is a 2-D tensor. When
931 932 933 934
                               `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
935
        soft_label (bool): a flag indicating whether to
936 937
                                           interpretate the given labels as soft
                                           labels, default `False`.
Y
Yibing Liu 已提交
938 939 940 941 942

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

    Raises:
943 944 945 946 947
        `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
948 949 950 951 952 953

    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
954
    """
F
fengjiayi 已提交
955
    helper = LayerHelper('cross_entropy', **locals())
Y
Yu Yang 已提交
956 957 958 959 960 961
    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
962
        attrs={"soft_label": soft_label})
Y
Yu Yang 已提交
963 964 965
    return out


F
fengjiayi 已提交
966
def square_error_cost(input, label):
Y
Yu Yang 已提交
967
    """
968 969
    **Square error cost layer**

970 971
    This layer accepts input predictions and target label and returns the
    squared error cost.
Y
ying 已提交
972

973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989
    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
990
        Variable: The tensor variable storing the element-wise squared error \
991
                  difference of input and label.
992 993 994 995 996 997 998 999

    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
1000
    """
F
fengjiayi 已提交
1001
    helper = LayerHelper('square_error_cost', **locals())
Y
Yu Yang 已提交
1002 1003 1004 1005 1006 1007 1008 1009 1010
    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
1011 1012
        type='square', inputs={'X': [minus_out]},
        outputs={'Out': [square_out]})
Y
Yu Yang 已提交
1013 1014 1015 1016 1017 1018 1019
    return square_out


def chunk_eval(input,
               label,
               chunk_scheme,
               num_chunk_types,
F
fengjiayi 已提交
1020
               excluded_chunk_types=None):
Y
Yu Yang 已提交
1021
    """
Y
yangyaming 已提交
1022
    This function computes and outputs the precision, recall and
1023
    F1-score of chunk detection.
Y
Yu Yang 已提交
1024
    """
F
fengjiayi 已提交
1025
    helper = LayerHelper("chunk_eval", **locals())
Y
Yu Yang 已提交
1026 1027 1028 1029 1030

    # prepare output
    precision = helper.create_tmp_variable(dtype="float32")
    recall = helper.create_tmp_variable(dtype="float32")
    f1_score = helper.create_tmp_variable(dtype="float32")
1031 1032 1033
    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
1034 1035 1036 1037 1038 1039 1040 1041

    helper.append_op(
        type="chunk_eval",
        inputs={"Inference": [input],
                "Label": [label]},
        outputs={
            "Precision": [precision],
            "Recall": [recall],
1042 1043 1044 1045
            "F1-Score": [f1_score],
            "NumInferChunks": [num_infer_chunks],
            "NumLabelChunks": [num_label_chunks],
            "NumCorrectChunks": [num_correct_chunks]
Y
Yu Yang 已提交
1046 1047 1048
        },
        attrs={
            "num_chunk_types": num_chunk_types,
G
guosheng 已提交
1049 1050
            "chunk_scheme": chunk_scheme,
            "excluded_chunk_types": excluded_chunk_types or []
Y
Yu Yang 已提交
1051
        })
1052 1053
    return (precision, recall, f1_score, num_infer_chunks, num_label_chunks,
            num_correct_chunks)
Y
Yu Yang 已提交
1054 1055 1056 1057 1058 1059 1060 1061 1062


def sequence_conv(input,
                  num_filters,
                  filter_size=3,
                  filter_stride=1,
                  padding=None,
                  bias_attr=None,
                  param_attr=None,
1063
                  act=None):
Y
Yu Yang 已提交
1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096
    """
    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)


1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120
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


def softmax(input, param_attr=None, bias_attr=None, use_cudnn=True):
    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
1121 1122 1123
def conv2d(input,
           num_filters,
           filter_size,
C
chengduoZH 已提交
1124 1125
           stride=1,
           padding=0,
1126
           dilation=1,
Y
Yu Yang 已提交
1127 1128 1129
           groups=None,
           param_attr=None,
           bias_attr=None,
C
chengduoZH 已提交
1130
           use_cudnn=True,
1131
           use_mkldnn=False,
1132 1133
           act=None,
           name=None):
Y
Yu Yang 已提交
1134
    """
C
chengduoZH 已提交
1135 1136 1137
    **Convlution2D Layer**

    The convolution2D layer calculates the output based on the input, filter
1138 1139 1140
    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
1141 1142
    The details of convolution layer, please refer UFLDL's `convolution,
    <http://ufldl.stanford.edu/tutorial/supervised/FeatureExtractionUsingConvolution/>`_ .
1143 1144 1145
    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
1146

1147
    For each input :math:`X`, the equation is:
C
refine  
chengduoZH 已提交
1148

C
chengduoZH 已提交
1149 1150
    .. math::

C
refine  
chengduoZH 已提交
1151
        Out = \sigma (W \\ast X + b)
C
chengduoZH 已提交
1152

C
chengduoZH 已提交
1153
    In the above equation:
C
chengduoZH 已提交
1154

1155 1156 1157 1158 1159
    * :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.
1160 1161
    * :math:`Out`: Output value, the shape of :math:`Out` and :math:`X` may be
                   different.
C
chengduoZH 已提交
1162 1163 1164

    Example:

1165 1166 1167
        - Input:

          Input shape: $(N, C_{in}, H_{in}, W_{in})$
C
refine  
chengduoZH 已提交
1168

1169
          Filter shape: $(C_{out}, C_{in}, H_f, W_f)$
C
refine  
chengduoZH 已提交
1170

1171 1172
        - Output:
          Output shape: $(N, C_{out}, H_{out}, W_{out})$
C
refine  
chengduoZH 已提交
1173

C
chengduoZH 已提交
1174
        Where
1175 1176

        .. math::
C
chengduoZH 已提交
1177

C
chengduoZH 已提交
1178 1179
        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
1180 1181

    Args:
1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193
       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.
1194 1195 1196
       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.
1197 1198 1199 1200 1201 1202 1203 1204 1205 1206
       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
1207 1208
       name(str|None): A name for this layer(optional). If set None, the layer
           will be named automatically.
C
chengduoZH 已提交
1209 1210

    Returns:
G
guosheng 已提交
1211
        Variable: The tensor variable storing the convolution and \
C
chengduoZH 已提交
1212 1213
                  non-linearity activation result.

C
refine  
chengduoZH 已提交
1214
    Raises:
1215 1216
        ValueError: If the shapes of input, filter_size, stride, padding and
                    groups mismatch.
C
refine  
chengduoZH 已提交
1217

C
chengduoZH 已提交
1218 1219 1220
    Examples:
        .. code-block:: python

1221 1222 1223 1224
          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
1225 1226 1227 1228 1229
    """
    if stride is None:
        stride = [1, 1]

    num_channels = input.shape[1]
1230 1231

    l_type = 'conv2d'
X
xzl 已提交
1232 1233
    if (num_channels == groups and num_filters % num_channels == 0 and
            not use_cudnn):
1234
        l_type = 'depthwise_conv2d'
1235 1236 1237 1238

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

Y
Yu Yang 已提交
1239 1240 1241 1242 1243 1244 1245
    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
1246 1247 1248
    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')
1249
    dilation = utils.convert_to_list(dilation, 2, 'dilation')
C
chengduoZH 已提交
1250

C
chengduoZH 已提交
1251 1252
    if not isinstance(use_cudnn, bool):
        raise ValueError("use_cudnn should be True or False")
Y
Yu Yang 已提交
1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269

    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(
1270
        type=l_type,
Y
Yu Yang 已提交
1271 1272 1273 1274 1275
        inputs={
            'Input': input,
            'Filter': filter_param,
        },
        outputs={"Output": pre_bias},
C
chengduoZH 已提交
1276 1277 1278
        attrs={
            'strides': stride,
            'paddings': padding,
1279
            'dilations': dilation,
C
chengduoZH 已提交
1280
            'groups': groups,
1281 1282
            'use_cudnn': use_cudnn,
            'use_mkldnn': use_mkldnn
C
chengduoZH 已提交
1283
        })
Y
Yu Yang 已提交
1284 1285 1286 1287 1288 1289

    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
1290
def sequence_pool(input, pool_type):
Y
Yu Yang 已提交
1291
    """
Y
yangyaming 已提交
1292 1293 1294
    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
1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319

    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)
F
fengjiayi 已提交
1320

L
Luo Tao 已提交
1321 1322
    Args:
        input(variable): The input variable which is a LoDTensor.
Y
yangyaming 已提交
1323
        pool_type (string): The pooling type of sequence_pool.
L
Luo Tao 已提交
1324 1325 1326 1327 1328 1329 1330 1331
            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
1332

Y
yangyaming 已提交
1333
             x = fluid.layers.data(name='x', shape=[7, 1],
L
Luo Tao 已提交
1334 1335 1336 1337 1338
                              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')
Y
Yu Yang 已提交
1339
    """
F
fengjiayi 已提交
1340
    helper = LayerHelper('sequence_pool', **locals())
Y
Yu Yang 已提交
1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351
    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
1352 1353 1354 1355 1356
    # 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
1357 1358 1359
    return pool_out


F
fengjiayi 已提交
1360
def sequence_first_step(input):
L
Luo Tao 已提交
1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374
    """
    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
1375

L
Luo Tao 已提交
1376 1377 1378 1379 1380 1381 1382 1383 1384
    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
1385

Y
yangyaming 已提交
1386
             x = fluid.layers.data(name='x', shape=[7, 1],
L
Luo Tao 已提交
1387 1388 1389
                              dtype='float32', lod_level=1)
             x_first_step = fluid.layers.sequence_first_step(input=x)
    """
1390 1391 1392
    return sequence_pool(input=input, pool_type="first")


F
fengjiayi 已提交
1393
def sequence_last_step(input):
L
Luo Tao 已提交
1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407
    """
    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
1408

L
Luo Tao 已提交
1409 1410 1411 1412 1413 1414 1415 1416 1417
    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
1418

Y
yangyaming 已提交
1419
             x = fluid.layers.data(name='x', shape=[7, 1],
L
Luo Tao 已提交
1420 1421 1422
                              dtype='float32', lod_level=1)
             x_last_step = fluid.layers.sequence_last_step(input=x)
    """
1423 1424 1425
    return sequence_pool(input=input, pool_type="last")


Y
Yu Yang 已提交
1426
def pool2d(input,
C
chengduoZH 已提交
1427 1428
           pool_size=-1,
           pool_type="max",
C
chengduoZH 已提交
1429 1430
           pool_stride=1,
           pool_padding=0,
C
caoying03 已提交
1431
           global_pooling=False,
C
chengduoZH 已提交
1432
           use_cudnn=True,
1433
           ceil_mode=False,
1434
           use_mkldnn=False,
C
caoying03 已提交
1435
           name=None):
Y
Yu Yang 已提交
1436 1437 1438 1439 1440 1441 1442 1443
    """
    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
1444

C
chengduoZH 已提交
1445 1446 1447 1448 1449
    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
1450 1451 1452 1453
    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
1454 1455
    if not isinstance(use_cudnn, bool):
        raise ValueError("use_cudnn should be True or False")
Y
Yu Yang 已提交
1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469

    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
1470
            "paddings": pool_padding,
1471
            "use_cudnn": use_cudnn,
1472 1473
            "ceil_mode": ceil_mode,
            "use_mkldnn": use_mkldnn
Y
Yu Yang 已提交
1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485
        })

    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
1486
               data_layout='NCHW',
1487 1488 1489
               name=None,
               moving_mean_name=None,
               moving_variance_name=None):
Y
Yu Yang 已提交
1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515
    """
    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(
1516
        attr=helper.bias_attr, shape=param_shape, dtype=dtype, is_bias=True)
Y
Yu Yang 已提交
1517

1518 1519 1520
    mean = helper.create_parameter(
        attr=ParamAttr(
            name=moving_mean_name, initializer=Constant(0.0), trainable=False),
Q
QI JUN 已提交
1521
        shape=param_shape,
1522 1523 1524 1525 1526 1527 1528 1529
        dtype=input.dtype)
    mean.stop_gradient = True

    variance = helper.create_parameter(
        attr=ParamAttr(
            name=moving_variance_name,
            initializer=Constant(1.0),
            trainable=False),
Q
QI JUN 已提交
1530
        shape=param_shape,
1531 1532
        dtype=input.dtype)
    variance.stop_gradient = True
Y
Yu Yang 已提交
1533 1534 1535 1536 1537 1538

    # 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
1539 1540
    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
1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566

    batch_norm_out = helper.create_tmp_variable(dtype)

    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
        },
        attrs={"momentum": momentum,
               "epsilon": epsilon,
               "is_test": is_test})

    return helper.append_activation(batch_norm_out)


G
guosheng 已提交
1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578
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**

1579
    Assume feature vectors exist on dimensions
G
guosheng 已提交
1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599
    :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.
1600
        scale(bool): Whether to learn the adaptive gain :math:`g` after
G
guosheng 已提交
1601
            normalization.
1602
        shift(bool): Whether to learn the adaptive bias :math:`b` after
G
guosheng 已提交
1603
            normalization.
1604
        begin_norm_axis(bool): The normalization will be performed along
G
guosheng 已提交
1605
            dimensions from :attr:`begin_norm_axis` to :attr:`rank(input)`.
1606
        epsilon(float): The small value added to the variance to prevent
G
guosheng 已提交
1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637
            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
1638
    if shift:
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
        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
1663
def beam_search_decode(ids, scores, name=None):
Y
Yu Yang 已提交
1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683
    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
1684 1685 1686
                     padding=0,
                     stride=1,
                     dilation=1,
C
caoying03 已提交
1687
                     param_attr=None,
1688
                     bias_attr=None,
C
chengduoZH 已提交
1689
                     use_cudnn=True,
1690
                     act=None,
C
caoying03 已提交
1691
                     name=None):
Y
Yu Yang 已提交
1692
    """
1693 1694 1695 1696 1697 1698 1699 1700
    **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
1701 1702
    layer, please refer to the following explanation and references
    `therein <http://www.matthewzeiler.com/wp-content/uploads/2017/07/cvpr2010.pdf>`_.
1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714

    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.
1715 1716
    * :math:`Out`: Output value, the shape of :math:`Out` and :math:`X` may be
                   different.
Y
Yu Yang 已提交
1717

1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730
    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
1731

1732 1733 1734 1735
        .. 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
1736 1737

    Args:
1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756
       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.
1757 1758
       param_attr(ParamAttr): The parameters to the Conv2d_transpose Layer.
                              Default: None
1759
       bias_attr(ParamAttr): Bias parameter for the Conv2d layer. Default: None
1760 1761
       use_cudnn(bool): Use cudnn kernel or not, it is valid only when the cudnn
           library is installed. Default: True
1762
       act(str): Activation type. Default: None
1763 1764
       name(str|None): A name for this layer(optional). If set None, the layer
           will be named automatically.
Y
Yu Yang 已提交
1765 1766

    Returns:
1767 1768 1769
       Variable: The tensor variable storing the convolution transpose result.

    Raises:
1770 1771
       ValueError: If the shapes of input, filter_size, stride, padding and
                   groups mismatch.
1772 1773 1774 1775

    Examples:
       .. code-block:: python

1776 1777 1778 1779
          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
1780 1781 1782 1783 1784 1785
    """
    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
1786 1787 1788
    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
1789

C
chengduoZH 已提交
1790 1791 1792
    if not isinstance(use_cudnn, bool):
        raise ValueError("use_cudnn should be True or False")

Y
Yu Yang 已提交
1793 1794 1795 1796 1797 1798 1799 1800
    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
1801 1802 1803 1804 1805

        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
1806
        filter_size = [filter_size_h, filter_size_w]
C
chengduoZH 已提交
1807 1808 1809
    else:
        filter_size = utils.convert_to_list(filter_size, 2,
                                            'conv2d_transpose.filter_size')
Y
Yu Yang 已提交
1810 1811 1812 1813 1814

    filter_shape = [input_channel, num_filters] + filter_size
    img_filter = helper.create_parameter(
        dtype=input.dtype, shape=filter_shape, attr=helper.param_attr)

1815
    pre_bias = helper.create_tmp_variable(dtype=input.dtype)
Y
Yu Yang 已提交
1816 1817 1818 1819
    helper.append_op(
        type='conv2d_transpose',
        inputs={'Input': [input],
                'Filter': [img_filter]},
1820
        outputs={'Output': pre_bias},
C
chengduoZH 已提交
1821 1822 1823 1824 1825 1826
        attrs={
            'strides': stride,
            'paddings': padding,
            'dilations': dilation,
            'use_cudnn': use_cudnn
        })
Y
Yu Yang 已提交
1827

1828 1829
    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
1830
    return out
Y
yangyaming 已提交
1831 1832


Y
yangyaming 已提交
1833
def sequence_expand(x, y, ref_level=-1, name=None):
1834
    """Sequence Expand Layer. This layer will expand the input variable **x**
Y
yangyaming 已提交
1835 1836 1837 1838
    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:
1839 1840 1841 1842 1843

    .. code-block:: text

        * Case 1
            x is a LoDTensor:
Y
yangyaming 已提交
1844 1845
                x.lod  = [[0,   2,        4]]
                x.data = [[a], [b], [c], [d]]
1846 1847 1848 1849 1850 1851
                x.dims = [4, 1]

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

Y
yangyaming 已提交
1852
            ref_level: 0
1853

Y
yangyaming 已提交
1854 1855 1856
            then output is a 1-level LoDTensor:
                out.lod =  [[0,   2,        4,        6,        8]]
                out.data = [[a], [b], [a], [b], [c], [d], [c], [d]]
1857 1858 1859 1860
                out.dims = [8, 1]

        * Case 2
            x is a Tensor:
Y
yangyaming 已提交
1861
                x.data = [[a], [b], [c]]
1862 1863 1864
                x.dims = [3, 1]

            y is a LoDTensor:
Y
yangyaming 已提交
1865
                y.lod = [[0, 2, 2, 5]]
1866

Y
yangyaming 已提交
1867
            ref_level: -1
1868

Y
yangyaming 已提交
1869 1870 1871
            then output is a Tensor:
                out.data = [[a], [a], [c], [c], [c]]
                out.dims = [5, 1]
1872 1873 1874
    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
1875 1876
        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
1877
        name(str|None): A name for this layer(optional). If set None, the layer
Y
yangyaming 已提交
1878
                        will be named automatically.
1879 1880 1881 1882 1883 1884 1885 1886 1887 1888

    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
1889
            out = layers.sequence_expand(x=x, y=y, ref_level=0)
1890
    """
Y
yangyaming 已提交
1891
    helper = LayerHelper('sequence_expand', input=x, **locals())
1892 1893 1894
    dtype = helper.input_dtype()
    tmp = helper.create_tmp_variable(dtype)
    helper.append_op(
Y
yangyaming 已提交
1895 1896 1897 1898 1899
        type='sequence_expand',
        inputs={'X': x,
                'Y': y},
        outputs={'Out': tmp},
        attrs={'ref_level': ref_level})
1900
    return tmp
1901 1902


Q
Qiao Longfei 已提交
1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934
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
1935 1936 1937 1938
def lstm_unit(x_t,
              hidden_t_prev,
              cell_t_prev,
              forget_bias=0.0,
Y
yangyaming 已提交
1939
              param_attr=None,
C
caoying03 已提交
1940 1941
              bias_attr=None,
              name=None):
Y
yangyaming 已提交
1942 1943 1944 1945
    """Lstm unit layer. The equation of a lstm step is:

        .. math::

1946
            i_t & = \sigma(W_{x_i}x_{t} + W_{h_i}h_{t-1} + b_i)
Y
yangyaming 已提交
1947

1948
            f_t & = \sigma(W_{x_f}x_{t} + W_{h_f}h_{t-1} + b_f)
Y
yangyaming 已提交
1949

1950
            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
1951

1952
            o_t & = \sigma(W_{x_o}x_{t} + W_{h_o}h_{t-1} + b_o)
Y
yangyaming 已提交
1953 1954 1955

            h_t & = o_t tanh(c_t)

1956 1957 1958 1959 1960 1961
    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
1962 1963 1964

        .. math::

1965
            L_{i_t} = W_{x_i}x_{t} + W_{h_i}h_{t-1} + b_i
Y
yangyaming 已提交
1966 1967 1968 1969 1970 1971 1972 1973

    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
1974
    This layer has two outputs including :math:`h_t` and :math:`o_t`.
Y
yangyaming 已提交
1975 1976

    Args:
Y
yangyaming 已提交
1977 1978 1979 1980 1981 1982
        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
1983
        forget_bias (float): The forget bias of lstm unit.
Y
yangyaming 已提交
1984 1985
        param_attr (ParamAttr): The attributes of parameter weights, used to set
            initializer, name etc.
Y
yangyaming 已提交
1986 1987
        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
1988 1989
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
Y
yangyaming 已提交
1990 1991

    Returns:
Y
yangyaming 已提交
1992
        tuple: The hidden value and cell value of lstm unit.
Y
yangyaming 已提交
1993 1994

    Raises:
1995 1996 1997 1998
        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
1999 2000 2001 2002 2003 2004

    Examples:

        .. code-block:: python

             x_t = fluid.layers.fc(input=x_t_data, size=10)
2005
             prev_hidden = fluid.layers.fc(input=prev_hidden_data, size=30)
Y
yangyaming 已提交
2006
             prev_cell = fluid.layers.fc(input=prev_cell_data, size=30)
Y
yangyaming 已提交
2007
             hidden_value, cell_value = fluid.layers.lstm_unit(x_t=x_t,
Y
yangyaming 已提交
2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023
                                                    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
2024
        raise ValueError("The 1st dimensions of x_t, hidden_t_prev and "
2025 2026 2027 2028
                         "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
2029 2030
                         "cell_t_prev must be the same.")

Y
yangyaming 已提交
2031 2032 2033
    if bias_attr is None:
        bias_attr = ParamAttr()

Y
yangyaming 已提交
2034
    size = cell_t_prev.shape[1]
2035
    concat_out = concat(input=[x_t, hidden_t_prev], axis=1)
Y
yangyaming 已提交
2036 2037
    fc_out = fc(input=concat_out,
                size=4 * size,
Y
yangyaming 已提交
2038
                param_attr=param_attr,
2039
                bias_attr=bias_attr)
Y
yangyaming 已提交
2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051
    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
2052
    return h, c
G
guosheng 已提交
2053 2054


C
caoying03 已提交
2055
def reduce_sum(input, dim=None, keep_dim=False, name=None):
G
guosheng 已提交
2056
    """
Y
yangyaming 已提交
2057
    Computes the sum of tensor elements over the given dimension.
G
guosheng 已提交
2058 2059 2060

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
Y
yangyaming 已提交
2061 2062 2063 2064
        dim (int|None): The dimension along which the sum is performed. If
            :attr:`None`, sum 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 :math:`dim < 0`,
G
guosheng 已提交
2065
            the dimension to reduce is :math:`rank + dim`.
2066
        keep_dim (bool|False): Whether to reserve the reduced dimension in the
Y
yangyaming 已提交
2067
            output Tensor. The result tensor will have one fewer dimension
G
guosheng 已提交
2068
            than the :attr:`input` unless :attr:`keep_dim` is true.
C
caoying03 已提交
2069 2070
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
G
guosheng 已提交
2071 2072 2073

    Returns:
        Variable: The reduced Tensor variable.
F
fengjiayi 已提交
2074

G
guosheng 已提交
2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098
    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]]
    """
    helper = LayerHelper('reduce_sum', **locals())
    out = helper.create_tmp_variable(dtype=helper.input_dtype())
    helper.append_op(
        type='reduce_sum',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
            'dim': dim if dim != None else 0,
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out
G
guosheng 已提交
2099 2100


C
caoying03 已提交
2101
def reduce_mean(input, dim=None, keep_dim=False, name=None):
G
guosheng 已提交
2102
    """
Y
yangyaming 已提交
2103
    Computes the mean of tensor elements over the given dimension.
G
guosheng 已提交
2104 2105 2106

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
Y
yangyaming 已提交
2107 2108 2109 2110
        dim (int|None): The dimension along which the mean is computed. If
            :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
G
guosheng 已提交
2111
            :math:`dim < 0`, the dimension to reduce is :math:`rank + dim`.
Y
yangyaming 已提交
2112 2113
        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
2114
            than the :attr:`input` unless :attr:`keep_dim` is true.
C
caoying03 已提交
2115 2116
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
G
guosheng 已提交
2117 2118 2119

    Returns:
        Variable: The reduced Tensor variable.
F
fengjiayi 已提交
2120

G
guosheng 已提交
2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144
    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]
            fluid.layers.reduce_mean(x, dim=1, keep_dim=True)  # [[0.475], [0.4]]
    """
    helper = LayerHelper('reduce_mean', **locals())
    out = helper.create_tmp_variable(dtype=helper.input_dtype())
    helper.append_op(
        type='reduce_mean',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
            'dim': dim if dim != None else 0,
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out
2145 2146


C
caoying03 已提交
2147
def reduce_max(input, dim=None, keep_dim=False, name=None):
2148
    """
Y
yangyaming 已提交
2149
    Computes the maximum of tensor elements over the given dimension.
2150 2151 2152

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
Y
yangyaming 已提交
2153 2154 2155 2156
        dim (int|None): The dimension along which the maximum is computed.
            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))`.
2157
            If :math:`dim < 0`, the dimension to reduce is :math:`rank + dim`.
Y
yangyaming 已提交
2158 2159
        keep_dim (bool): Whether to reserve the reduced dimension in the
            output Tensor. The result tensor will have one fewer dimension
2160
            than the :attr:`input` unless :attr:`keep_dim` is true.
C
caoying03 已提交
2161 2162
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
2163 2164 2165

    Returns:
        Variable: The reduced Tensor variable.
Y
yangyaming 已提交
2166

2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192
    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]]
    """
    helper = LayerHelper('reduce_max', **locals())
    out = helper.create_tmp_variable(dtype=helper.input_dtype())
    helper.append_op(
        type='reduce_max',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
            'dim': dim if dim != None else 0,
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out


C
caoying03 已提交
2193
def reduce_min(input, dim=None, keep_dim=False, name=None):
2194
    """
Y
yangyaming 已提交
2195
    Computes the minimum of tensor elements over the given dimension.
2196 2197 2198

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
Y
yangyaming 已提交
2199 2200 2201 2202
        dim (int|None): The dimension along which the minimum is computed.
            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))`.
2203
            If :math:`dim < 0`, the dimension to reduce is :math:`rank + dim`.
Y
yangyaming 已提交
2204 2205
        keep_dim (bool): Whether to reserve the reduced dimension in the
            output Tensor. The result tensor will have one fewer dimension
2206
            than the :attr:`input` unless :attr:`keep_dim` is true.
C
caoying03 已提交
2207 2208
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
2209 2210 2211

    Returns:
        Variable: The reduced Tensor variable.
Y
yangyaming 已提交
2212

2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236
    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]]
    """
    helper = LayerHelper('reduce_min', **locals())
    out = helper.create_tmp_variable(dtype=helper.input_dtype())
    helper.append_op(
        type='reduce_min',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
            'dim': dim if dim != None else 0,
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out
G
guosheng 已提交
2237 2238


2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252
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.
        dim (int|None): The dimension along which the product is performed. If
            :attr:`None`, multipy 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 :math:`dim < 0`,
            the dimension to reduce is :math:`rank + dim`.
        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
2253
        name(str|None): A name for this layer(optional). If set None, the
Z
zhouhanqing 已提交
2254
            layer will be named automatically.
2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268

    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
2269
            fluid.layers.reduce_prod(x, dim=1,
Z
zhouhanqing 已提交
2270
                                     keep_dim=True)  # [[0.027], [0.0084]]
2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285
    """
    helper = LayerHelper('reduce_prod', **locals())
    out = helper.create_tmp_variable(dtype=helper.input_dtype())
    helper.append_op(
        type='reduce_prod',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
            'dim': dim if dim != None else 0,
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out


C
caoying03 已提交
2286
def split(input, num_or_sections, dim=-1, name=None):
G
guosheng 已提交
2287
    """
C
caoying03 已提交
2288
    Split the input tensor into multiple sub-tensors.
G
guosheng 已提交
2289 2290 2291

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
C
caoying03 已提交
2292 2293 2294 2295 2296
        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
2297
            :attr:`dim` dimension orderly.
C
caoying03 已提交
2298
        dim (int): The dimension along which to split. If :math:`dim < 0`, the
G
guosheng 已提交
2299
            dimension to split along is :math:`rank(input) + dim`.
C
caoying03 已提交
2300 2301
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
G
guosheng 已提交
2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343

    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]
            x0, x1, x2 = fluid.layers.split(x, num_or_sections=[2, 3, 4], dim=1)
            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
2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376


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
2377
          normed = fluid.layers.l2_normalize(x=data, axis=1)
C
caoying03 已提交
2378 2379
    """

F
fengjiayi 已提交
2380 2381
    if len(x.shape) == 1:
        axis = 0
C
caoying03 已提交
2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407

    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={
            "dim": 1 if axis is None else axis,
            "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
2408
    # exanpsion is requred. Once a general broadcast rule is spported, this
C
caoying03 已提交
2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425
    # 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
2426 2427


2428
def matmul(x, y, transpose_x=False, transpose_y=False, name=None):
G
guosheng 已提交
2429
    """
Y
ying 已提交
2430 2431 2432 2433
    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
2434

C
chengduoZH 已提交
2435
    The actual behavior depends on the shapes of :math:`x`, :math:`y` and the
2436
    flag values of :attr:`transpose_x`, :attr:`transpose_y`. Specifically:
G
guosheng 已提交
2437

2438 2439 2440 2441 2442
    - 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
2443
      :math:`[1, D]` in transposed form.
G
guosheng 已提交
2444

C
chengduoZH 已提交
2445
    - After transpose, the two tensors are 2-D or n-D and matrix multiplication
2446
      performs in the following way.
G
guosheng 已提交
2447

2448
      - If both are 2-D, they are multiplied like conventional matrices.
C
chengduoZH 已提交
2449
      - If either is n-D, it is treated as a stack of matrices residing in the
Y
ying 已提交
2450
        last two dimensions and a batched matrix multiply supporting broadcast
2451
        applies on the two tensors.
G
guosheng 已提交
2452

Y
ying 已提交
2453 2454
    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
2455
    removed after matrix multiplication.
G
guosheng 已提交
2456 2457 2458

    Args:
        x (Variable): The input variable which is a Tensor or LoDTensor.
2459 2460 2461
        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.
2462
        name(str|None): A name for this layer(optional). If set None, the layer
2463
            will be named automatically.
G
guosheng 已提交
2464 2465

    Returns:
2466
        Variable: The product Tensor variable.
G
guosheng 已提交
2467

G
guosheng 已提交
2468 2469 2470
    Examples:
        .. code-block:: python

2471
            # Examples to clarify shapes of the inputs and output
C
chengduoZH 已提交
2472 2473
            # x: [B, ..., M, K], y: [B, ..., K, N]
            fluid.layers.matmul(x, y)  # out: [B, ..., M, N]
Y
ying 已提交
2474

2475 2476
            # x: [B, M, K], y: [B, K, N]
            fluid.layers.matmul(x, y)  # out: [B, M, N]
Y
ying 已提交
2477

2478 2479
            # x: [B, M, K], y: [K, N]
            fluid.layers.matmul(x, y)  # out: [B, M, N]
Y
ying 已提交
2480

2481 2482
            # x: [M, K], y: [K, N]
            fluid.layers.matmul(x, y)  # out: [M, N]
Y
ying 已提交
2483 2484 2485 2486

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

2487 2488
            # x: [K], y: [K]
            fluid.layers.matmul(x, y)  # out: [1]
2489

Y
ying 已提交
2490
            # x: [M], y: [N]
2491
            fluid.layers.matmul(x, y, True, True)  # out: [M, N]
G
guosheng 已提交
2492
    """
Y
ying 已提交
2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504

    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
2505
            y_shape = y_shape + [1]
Y
ying 已提交
2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521

        # 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)

2522
    helper = LayerHelper('matmul', **locals())
Y
ying 已提交
2523
    out = helper.create_tmp_variable(dtype=x.dtype)
G
guosheng 已提交
2524
    helper.append_op(
2525 2526 2527 2528 2529 2530 2531
        type='matmul',
        inputs={'X': x,
                'Y': y},
        outputs={'Out': out},
        attrs={'transpose_X': transpose_x,
               'transpose_Y': transpose_y})
    return out
2532 2533


W
wanghaoshuang 已提交
2534
def edit_distance(input, label, normalized=True, ignored_tokens=None,
W
wanghaoshuang 已提交
2535
                  name=None):
2536
    """
Y
ying 已提交
2537 2538 2539 2540 2541 2542 2543 2544 2545
    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
2546

Y
ying 已提交
2547
    "kitten" -> "sitten" -> "sittin" -> "sitting"
W
wanghaoshuang 已提交
2548

Y
ying 已提交
2549 2550 2551 2552
    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
2553

Y
ying 已提交
2554 2555 2556
    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
2557

2558 2559 2560 2561 2562
    Args:

        input(Variable): The indices for hypothesis strings.

        label(Variable): The indices for reference strings.
W
wanghaoshuang 已提交
2563

Y
ying 已提交
2564 2565
        normalized(bool): Indicated whether to normalize the edit distance by
                          the length of reference string.
2566

Y
ying 已提交
2567 2568
        ignored_tokens(list of int): Tokens that should be removed before
                                     calculating edit distance.
2569

W
wanghaoshuang 已提交
2570
    Returns:
W
wanghaoshuang 已提交
2571
        Variable: sequence-to-sequence edit distance in shape [batch_size, 1].
W
wanghaoshuang 已提交
2572 2573 2574 2575 2576

    Examples:
        .. code-block:: python

            x = fluid.layers.data(name='x', shape=[8], dtype='float32')
2577 2578
            y = fluid.layers.data(name='y', shape=[7], dtype='float32')

2579
            cost = fluid.layers.edit_distance(input=x,label=y)
2580
    """
2581
    helper = LayerHelper("edit_distance", **locals())
2582

2583
    # remove some tokens from input and labels
W
wanghaoshuang 已提交
2584
    if ignored_tokens is not None and len(ignored_tokens) > 0:
2585 2586 2587 2588 2589 2590 2591
        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
2592
            attrs={"tokens": ignored_tokens})
2593 2594 2595 2596 2597 2598
        input = erased_input

        helper.append_op(
            type="sequence_erase",
            inputs={"X": [label]},
            outputs={"Out": [erase_label]},
W
wanghaoshuang 已提交
2599
            attrs={"tokens": ignored_tokens})
2600 2601
        label = erased_label

2602 2603
    # edit distance op
    edit_distance_out = helper.create_tmp_variable(dtype="int64")
2604
    sequence_num = helper.create_tmp_variable(dtype="int64")
2605 2606 2607 2608
    helper.append_op(
        type="edit_distance",
        inputs={"Hyps": [input],
                "Refs": [label]},
2609 2610
        outputs={"Out": [edit_distance_out],
                 "SequenceNum": [sequence_num]},
2611 2612
        attrs={"normalized": normalized})

2613
    return edit_distance_out, sequence_num
2614 2615 2616 2617 2618


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
2619 2620 2621 2622
    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.
2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651

    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
2652 2653 2654 2655 2656 2657
        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).
2658

Y
ying 已提交
2659 2660 2661
        blank(int): the blank label index of Connectionist Temporal
                    Classification (CTC) loss, which is in thehalf-opened
                    interval [0, num_classes + 1).
2662 2663

    Returns:
2664
        Variable: CTC greedy decode result. If all the sequences in result were
2665
        empty, the result LoDTensor will be [-1] with LoD [[0]] and dims [1, 1].
2666 2667 2668 2669 2670

    Examples:
        .. code-block:: python

            x = fluid.layers.data(name='x', shape=[8], dtype='float32')
W
wanghaoshuang 已提交
2671

2672
            cost = fluid.layers.ctc_greedy_decoder(input=x, blank=0)
W
wanghaoshuang 已提交
2673
    """
2674
    helper = LayerHelper("ctc_greedy_decoder", **locals())
2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689
    # top 1 op
    topk_out = helper.create_tmp_variable(dtype=input.dtype)
    topk_indices = helper.create_tmp_variable(dtype="int64")
    helper.append_op(
        type="top_k",
        inputs={"X": [input]},
        outputs={"Out": [topk_out],
                 "Indices": [topk_indices]},
        attrs={"k": 1})

    # 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
2690
        outputs={"Output": [ctc_out]},
2691 2692
        attrs={"merge_repeated": True,
               "blank": blank})
2693
    return ctc_out
2694 2695


F
fengjiayi 已提交
2696
def warpctc(input, label, blank=0, norm_by_times=False):
W
wanghaoshuang 已提交
2697
    """
2698 2699
    An operator integrating the open source Warp-CTC library
    (https://github.com/baidu-research/warp-ctc)
W
wanghaoshuang 已提交
2700
    to compute Connectionist Temporal Classification (CTC) loss.
2701 2702
    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
2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715
    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.
2716
       blank: (int, default: 0), the blank label index of Connectionist
W
wanghaoshuang 已提交
2717 2718
         Temporal Classification (CTC) loss, which is in the
         half-opened interval [0, num_classes + 1).
2719
       norm_by_times: (bool, default: false), whether to normalize
W
wanghaoshuang 已提交
2720
       the gradients by the number of time-step, which is also the
2721 2722
       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
2723 2724

    Returns:
2725 2726
        Variable: The Connectionist Temporal Classification (CTC) loss,
        which is a 2-D Tensor of the shape [batch_size, 1].
W
wanghaoshuang 已提交
2727 2728 2729

    Examples:
        .. code-block:: python
2730 2731 2732 2733
            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
2734 2735 2736
            cost = layers.warpctc(input=y_predict, label=y)

    """
F
fengjiayi 已提交
2737
    helper = LayerHelper('warpctc', **locals())
W
wanghaoshuang 已提交
2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748
    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
2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 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 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802


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
2803 2804


2805
@autodoc()
Y
Yang Yu 已提交
2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831
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
2832 2833 2834 2835 2836 2837 2838 2839 2840
    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
2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856

    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
2857
    return cost / (num_neg_samples + 1)
2858 2859


Y
fix ci.  
ying 已提交
2860
def transpose(x, perm, name=None):
Y
ying 已提交
2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879
    """
    **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
2880
            x_transposed = layers.transpose(x, perm=[1, 0, 2])
Y
ying 已提交
2881 2882
    """

Y
fix ci.  
ying 已提交
2883
    if len(perm) != len(x.shape):
Y
ying 已提交
2884 2885 2886
        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
2887 2888 2889 2890 2891 2892
    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
2893 2894

    helper = LayerHelper('transpose', **locals())
Y
fix ci.  
ying 已提交
2895
    out = helper.create_tmp_variable(x.dtype)
Y
ying 已提交
2896 2897
    helper.append_op(
        type='transpose',
Y
fix ci.  
ying 已提交
2898
        inputs={'X': [x]},
Y
ying 已提交
2899 2900 2901
        outputs={'Out': [out]},
        attrs={'axis': perm})
    return out
2902 2903


2904
def im2sequence(input, filter_size=1, stride=1, padding=0, name=None):
2905
    """
2906 2907 2908 2909 2910 2911 2912
    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:
2913 2914 2915 2916 2917 2918 2919 2920 2921 2922

    .. 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
2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940

        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.

2941 2942 2943
        name (int): The name of this layer. It is optional.

    Returns:
W
wanghaoshuang 已提交
2944 2945 2946 2947 2948
        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.
2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977

    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
2978 2979 2980
            filter = [2, 2]
            stride = [1, 1]
            padding = [0, 0]
2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000

            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

3001 3002
            output = fluid.layers.im2sequence(
                input=layer, stride=[1, 1], filter_size=[2, 2])
3003 3004

    """
W
wanghaoshuang 已提交
3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015

    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])

3016
    helper = LayerHelper('im2sequence', **locals())
3017 3018
    out = helper.create_tmp_variable(dtype=helper.input_dtype())
    helper.append_op(
3019
        type='im2sequence',
3020 3021 3022
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
W
wanghaoshuang 已提交
3023 3024 3025
            'kernels': filter_size,
            'strides': stride,
            'paddings': padding,
3026 3027
        })
    return out
3028 3029


3030 3031 3032 3033
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
3034
    equation of row convolution is as follows:
3035 3036 3037 3038 3039 3040 3041

    .. 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
3042
    * :math:`\\tau`: Future context size.
3043 3044 3045 3046 3047 3048 3049 3050 3051 3052
    * :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
3053 3054
        future_context_size (int): Future context size. Please note, the shape
            of convolution kernel is [future_context_size + 1, D].
3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079
        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
3080
    return helper.append_activation(out)
3081 3082


3083 3084 3085 3086
def multiplex(inputs, index):
    """
    **Multiplex Layer**

Y
yangyaming 已提交
3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101
    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]`.
3102 3103

    Args:
Y
yangyaming 已提交
3104 3105
       inputs (list): A list of variables to gather from. All variables have the
                same shape and the rank is at least 2.
3106
       index (Variable): Tensor<int32>, index variable which is a 2-D tensor
Y
yangyaming 已提交
3107
                with shape [M, 1] where M is the batch size.
3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120

    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
3121 3122 3123 3124 3125 3126

    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)
3127 3128 3129 3130 3131 3132
    helper.append_op(
        type='multiplex',
        inputs={'X': inputs,
                'Ids': index},
        outputs={'Out': [out]})
    return out
3133 3134 3135 3136 3137


def softmax_with_cross_entropy(logits, label, soft_label=False):
    """
    **Softmax With Cross Entropy Operator.**
3138

3139 3140 3141 3142
    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.
3143

3144 3145 3146
    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.
3147

3148 3149 3150
    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.
3151

3152
    The equation is as follows:
3153

3154
    1) Hard label (one-hot label, so every sample has exactly one class)
3155

3156 3157 3158 3159
    .. math::

        loss_j =  -\\text{logit}_{label_j} +
        \\log\\left(\\sum_{i=0}^{K}\\exp(\\text{logit}_i)\\right), j = 1,..., K
3160

3161 3162 3163
    2) Soft label (each sample can have a distribution over all classes)

    .. math::
3164

3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208
        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)
            out = fluid.layers.softmax_with_cross_entropy(logits=fc, label=label)
    """
    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. **

    This operator computes the smooth l1 loss for X and Y.
    The operator takes the first dimension of X and Y as batch size.
    For each instance, it computes the smooth l1 loss element by element first
    and then sums all the losses. So the shape of Out is [batch_size, 1].
3209

3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234
    Args:
        x (Variable): A tensor with rank at least 2. The input value of smooth
            l1 loss op with shape [batch_size, dim1, ..., dimN].
        y (Variable): A tensor with rank at least 2. The target value of smooth
            l1 loss op with same shape as x.
        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,
            the out smooth l1 loss will be multiplied by this tensor element
            by element.
        sigma (float|None): Hyper parameter of smooth l1 loss op. A float scalar
            with default value 1.0.
    Returns:
        Variable: A tensor with rank be 2. The output smooth l1 loss with
            shape [batch_size, 1].

    Examples:
        .. code-block:: python

            data = fluid.layers.data(name='data', shape=[128], dtype='float32')
            label = fluid.layers.data(name='label', shape=[100], dtype='int64')
            fc = fluid.layers.fc(input=data, size=100)
F
fengjiayi 已提交
3235
            out = fluid.layers.smooth_l1(x=fc, y=label)
3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251
    """
    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
3252 3253 3254 3255 3256 3257 3258 3259 3260


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
3261
        input(variable):  A Tensor/LodTensor of indices, last dimension must be 1.
3262 3263 3264 3265 3266 3267
        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
3268 3269
        .. code-block:: python

3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290
        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
3291 3292


Y
Yu Yang 已提交
3293
def autoincreased_step_counter(counter_name=None, begin=1, step=1):
Y
Yu Yang 已提交
3294
    """
Y
Yu Yang 已提交
3295
    NOTE: The counter will be automatically increased by 1 every mini-batch
Y
Yu Yang 已提交
3296
    Return the run counter of the main program, which is started with 1.
Y
Yu Yang 已提交
3297 3298 3299 3300 3301 3302

    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
3303 3304 3305
    Returns(Variable): The global run counter.
    """
    helper = LayerHelper('global_step_counter')
Y
Yu Yang 已提交
3306 3307
    if counter_name is None:
        counter_name = '@STEP_COUNTER@'
Y
Yu Yang 已提交
3308 3309 3310 3311 3312
    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
3313
                value=begin - 1, force_cpu=True))
Y
Yu Yang 已提交
3314 3315 3316
        helper.main_program.global_block().prepend_op(
            type='increment',
            inputs={'X': [counter]},
Y
Yu Yang 已提交
3317 3318
            outputs={'Out': [counter]},
            attrs={'step': float(step)})
Y
Yu Yang 已提交
3319 3320 3321
        counter.stop_gradient = True

    return counter
C
caoying03 已提交
3322 3323


3324
def reshape(x, shape, actual_shape=None, act=None, inplace=True, name=None):
C
caoying03 已提交
3325
    """
C
caoying03 已提交
3326 3327
    Gives a new shape to the input Tensor without changing its data.

3328 3329 3330 3331 3332
    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
3333

3334
    Some tricks exist when specifying the target shape.
C
caoying03 已提交
3335

3336 3337 3338 3339 3340 3341 3342 3343 3344
    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.

    1. 0 means the actual dimension value is going to be copied from the
    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
3345 3346

    1. Given a 3-D tensor x with a shape [2, 4, 6], and the target shape
3347 3348
    is [6, 8], the reshape operator will transform x into a 2-D tensor with 
    shape [6, 8] and leaving x's data unchanged.
C
caoying03 已提交
3349 3350

    1. Given a 3-D tensor x with a shape [2, 4, 6], and the target shape
3351 3352 3353 3354 3355
    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
    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 
    dimensions.
C
caoying03 已提交
3356 3357

    1. Given a 3-D tensor x with a shape [2, 4, 6], and the target shape
3358 3359 3360 3361
    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
3362 3363 3364 3365 3366

    Args:
        input(variable): The input tensor.
        shape(list): The new shape. At most one dimension of the new shape can
                     be -1.
3367 3368 3369 3370 3371
        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
3372 3373 3374 3375 3376 3377 3378 3379 3380
        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
3381
            data = fluid.layers.data(
3382
                name='data', shape=[2, 4, 6], dtype='float32')
C
caoying03 已提交
3383
            reshaped = fluid.layers.reshape(
3384
                x=data, shape=[-1, 0, 3, 2], act='tanh', inplace=True)
C
caoying03 已提交
3385 3386 3387 3388 3389
    """

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

3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404
    # 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
3405 3406 3407 3408
    helper = LayerHelper("reshape", **locals())
    reshaped = helper.create_tmp_variable(dtype=x.dtype)
    helper.append_op(
        type="reshape",
3409 3410 3411
        inputs={"X": x,
                "Shape": actual_shape}
        if isinstance(actual_shape, Variable) else {"X": x},
C
caoying03 已提交
3412 3413 3414 3415 3416
        attrs={"shape": shape,
               "inplace": inplace},
        outputs={"Out": reshaped})

    return helper.append_activation(reshaped)
3417 3418


Y
yangyaming 已提交
3419
def lod_reset(x, y=None, target_lod=None):
Y
yangyaming 已提交
3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511
    """
    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
3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 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


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

          data = fluid.layers.data(name="data", shape=[3, 112, 112], dtype="float32")
          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