nn.py 304.0 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13
# Copyright (c) 2018 PaddlePaddle Authors. All Rights Reserved.
#
# 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
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
Y
Yu Yang 已提交
14
"""
15
All layers just related to the neural network.
Y
Yu Yang 已提交
16 17
"""

18 19
from __future__ import print_function

20
import numpy as np
P
peizhilin 已提交
21
import os
Y
Yu Yang 已提交
22 23
from ..layer_helper import LayerHelper
from ..initializer import Normal, Constant
S
sneaxiy 已提交
24
from ..framework import Variable, OpProtoHolder
Y
yangyaming 已提交
25
from ..param_attr import ParamAttr
S
sneaxiy 已提交
26
from .layer_function_generator import autodoc, templatedoc, _generate_doc_string_
27 28
from .tensor import concat
from . import utils
F
fengjiayi 已提交
29
from .. import unique_name
30
from functools import reduce
31
from .. import core
Y
Yu Yang 已提交
32 33

__all__ = [
X
Xin Pan 已提交
34 35
    'fc',
    'embedding',
P
peizhilin 已提交
36
    'dynamic_lstm',
X
Xin Pan 已提交
37 38 39 40
    'dynamic_lstmp',
    'dynamic_gru',
    'gru_unit',
    'linear_chain_crf',
P
peizhilin 已提交
41
    'crf_decoding',
X
Xin Pan 已提交
42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
    'cos_sim',
    'cross_entropy',
    'square_error_cost',
    'chunk_eval',
    'sequence_conv',
    'conv2d',
    'conv3d',
    'sequence_pool',
    'sequence_softmax',
    'softmax',
    'pool2d',
    'pool3d',
    'batch_norm',
    'beam_search_decode',
    'conv2d_transpose',
    'conv3d_transpose',
    'sequence_expand',
    'sequence_expand_as',
    'sequence_pad',
Y
Yibing Liu 已提交
61
    'sequence_unpad',
X
Xin Pan 已提交
62 63 64 65 66 67 68 69
    'lstm_unit',
    'reduce_sum',
    'reduce_mean',
    'reduce_max',
    'reduce_min',
    'reduce_prod',
    'sequence_first_step',
    'sequence_last_step',
Y
Yibing Liu 已提交
70
    'sequence_slice',
X
Xin Pan 已提交
71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99
    'dropout',
    'split',
    'ctc_greedy_decoder',
    'edit_distance',
    'l2_normalize',
    'matmul',
    'topk',
    'warpctc',
    'sequence_reshape',
    'transpose',
    'im2sequence',
    'nce',
    'hsigmoid',
    'beam_search',
    'row_conv',
    'multiplex',
    'layer_norm',
    'softmax_with_cross_entropy',
    'smooth_l1',
    'one_hot',
    'autoincreased_step_counter',
    'reshape',
    'squeeze',
    'unsqueeze',
    'lod_reset',
    'lrn',
    'pad',
    'pad_constant_like',
    'label_smooth',
P
peizhilin 已提交
100
    'roi_pool',
J
jerrywgz 已提交
101
    'roi_align',
X
Xin Pan 已提交
102 103 104 105
    'dice_loss',
    'image_resize',
    'image_resize_short',
    'resize_bilinear',
106
    'resize_nearest',
X
Xin Pan 已提交
107 108 109 110 111 112
    'gather',
    'scatter',
    'sequence_scatter',
    'random_crop',
    'mean_iou',
    'relu',
C
chengduo 已提交
113
    'selu',
X
Xin Pan 已提交
114 115 116
    'log',
    'crop',
    'rank_loss',
M
minqiyang 已提交
117
    'margin_rank_loss',
X
Xin Pan 已提交
118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160
    'elu',
    'relu6',
    'pow',
    'stanh',
    'hard_sigmoid',
    'swish',
    'prelu',
    'brelu',
    'leaky_relu',
    'soft_relu',
    'flatten',
    'sequence_mask',
    'stack',
    'pad2d',
    'unstack',
    'sequence_enumerate',
    'expand',
    'sequence_concat',
    'scale',
    'elementwise_add',
    'elementwise_div',
    'elementwise_sub',
    'elementwise_mul',
    'elementwise_max',
    'elementwise_min',
    'elementwise_pow',
    'uniform_random_batch_size_like',
    'gaussian_random',
    'sampling_id',
    'gaussian_random_batch_size_like',
    'sum',
    'slice',
    'shape',
    'logical_and',
    'logical_or',
    'logical_xor',
    'logical_not',
    'clip',
    'clip_by_norm',
    'mean',
    'mul',
    'sigmoid_cross_entropy_with_logits',
    'maxout',
J
JiabinYang 已提交
161
    'space_to_depth',
W
whs 已提交
162
    'affine_grid',
S
sneaxiy 已提交
163
    'sequence_reverse',
164
    'affine_channel',
B
barrierye 已提交
165
    'similarity_focus',
M
minqiyang 已提交
166
    'hash',
D
dengkaipeng 已提交
167
    'grid_sampler',
G
gmcather 已提交
168 169
    'log_loss',
    'add_position_encoding',
Q
Qiao Longfei 已提交
170
    'bilinear_tensor_product',
Y
Yu Yang 已提交
171 172 173 174 175 176 177 178 179
]


def fc(input,
       size,
       num_flatten_dims=1,
       param_attr=None,
       bias_attr=None,
       act=None,
J
Jacek Czaja 已提交
180
       is_test=False,
181
       name=None):
Y
Yu Yang 已提交
182
    """
183
    **Fully Connected Layer**
Y
Yu Yang 已提交
184

185 186 187 188 189 190 191 192
    This function creates a fully connected layer in the network. It can take
    multiple tensors as its inputs. It 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, if activation is not None, it will be applied
F
fengjiayi 已提交
193
    to the output as well.
C
caoying03 已提交
194

C
caoying03 已提交
195
    This process can be formulated as follows:
196 197 198

    .. math::

199
        Out = Act({\sum_{i=0}^{N-1}X_iW_i + b})
200 201 202

    In the above equation:

C
caoying03 已提交
203 204 205 206
    * :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).
207
    * :math:`Act`: The activation function.
C
caoying03 已提交
208
    * :math:`Out`: The output tensor.
Y
Yu Yang 已提交
209 210

    Args:
R
ranqiu 已提交
211 212 213 214 215 216 217 218 219 220 221 222 223 224 225
        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
226 227
            of this layer. If it is set to False, no bias will be added to the output units.
            If it is set to None, the bias is initialized zero. Default: None.
R
ranqiu 已提交
228
        act (str, default None): Activation to be applied to the output of this layer.
J
Jacek Czaja 已提交
229
        is_test(bool): A flag indicating whether execution is in test phase.
R
ranqiu 已提交
230
        name (str, default None): The name of this layer.
Y
Yu Yang 已提交
231

232
    Returns:
F
fengjiayi 已提交
233
        Variable: The transformation result.
234 235

    Raises:
C
caoying03 已提交
236
        ValueError: If rank of the input tensor is less than 2.
237 238 239 240

    Examples:
        .. code-block:: python

F
fengjiayi 已提交
241
          data = fluid.layers.data(name="data", shape=[32, 32], dtype="float32")
242
          fc = fluid.layers.fc(input=data, size=1000, act="tanh")
Y
Yu Yang 已提交
243
    """
C
caoying03 已提交
244

C
caoying03 已提交
245
    helper = LayerHelper("fc", **locals())
Y
Yu Yang 已提交
246 247 248 249

    dtype = helper.input_dtype()

    mul_results = []
250 251
    for input_var, param_attr in helper.iter_inputs_and_params():
        input_shape = input_var.shape
Y
Yu Yang 已提交
252 253 254
        param_shape = [
            reduce(lambda a, b: a * b, input_shape[num_flatten_dims:], 1)
        ] + [size]
Y
ying 已提交
255

Y
Yu Yang 已提交
256
        w = helper.create_parameter(
257
            attr=param_attr, shape=param_shape, dtype=dtype, is_bias=False)
X
Xin Pan 已提交
258
        tmp = helper.create_variable_for_type_inference(dtype)
259
        helper.append_op(
260 261 262
            type="mul",
            inputs={"X": input_var,
                    "Y": w},
263
            outputs={"Out": tmp},
M
mozga-intel 已提交
264 265
            attrs={"x_num_col_dims": num_flatten_dims,
                   "y_num_col_dims": 1})
266 267 268 269
        mul_results.append(tmp)

    if len(mul_results) == 1:
        pre_bias = mul_results[0]
270
    else:
X
Xin Pan 已提交
271
        pre_bias = helper.create_variable_for_type_inference(dtype)
272
        helper.append_op(
273 274 275
            type="sum",
            inputs={"X": mul_results},
            outputs={"Out": pre_bias},
X
Xin Pan 已提交
276
            attrs={"use_mkldnn": False})
277 278 279 280
    # add bias
    pre_activation = helper.append_bias_op(pre_bias, dim_start=num_flatten_dims)
    # add activation
    return helper.append_activation(pre_activation)
Y
Yu Yang 已提交
281 282


283 284 285
def embedding(input,
              size,
              is_sparse=False,
286
              is_distributed=False,
287 288 289
              padding_idx=None,
              param_attr=None,
              dtype='float32'):
Y
Yu Yang 已提交
290
    """
291 292
    **Embedding Layer**

293
    This layer is used to lookup embeddings of IDs, provided by :attr:`input`, in
294 295
    a lookup table. The result of this lookup is the embedding of each ID in the
    :attr:`input`.
296 297 298

    All the input variables are passed in as local variables to the LayerHelper
    constructor.
Y
Yu Yang 已提交
299 300

    Args:
301 302 303 304 305
        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.
306
        is_distributed(bool): Whether to run lookup table from remote parameter server.
307 308
        padding_idx(int|long|None): If :attr:`None`, it makes no effect to lookup.
            Otherwise the given :attr:`padding_idx` indicates padding the output
309
            with zeros whenever lookup encounters it in :attr:`input`. If
310
            :math:`padding_idx < 0`, the :attr:`padding_idx` to use in lookup is
311 312
            :math:`size[0] + dim`.
        param_attr(ParamAttr): Parameters for this layer
313
        dtype(np.dtype|core.VarDesc.VarType|str): The type of data : float32, float_16, int etc
Y
Yu Yang 已提交
314

315 316 317
    Returns:
        Variable: The tensor variable storing the embeddings of the \
                  supplied inputs.
Y
Yu Yang 已提交
318

319 320
    Examples:
        .. code-block:: python
Y
Yu Yang 已提交
321

C
chengduoZH 已提交
322
          dict_size = len(dataset.ids)
323
          data = fluid.layers.data(name='ids', shape=[32, 32], dtype='float32')
C
chengduoZH 已提交
324
          fc = fluid.layers.embedding(input=data, size=[dict_size, 16])
Y
Yu Yang 已提交
325 326 327 328 329
    """

    helper = LayerHelper('embedding', **locals())
    w = helper.create_parameter(
        attr=helper.param_attr, shape=size, dtype=dtype, is_bias=False)
X
Xin Pan 已提交
330
    tmp = helper.create_variable_for_type_inference(dtype)
331 332
    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
333 334 335 336 337
    helper.append_op(
        type='lookup_table',
        inputs={'Ids': input,
                'W': w},
        outputs={'Out': tmp},
338 339 340 341 342
        attrs={
            'is_sparse': is_sparse,
            'is_distributed': is_distributed,
            'padding_idx': padding_idx
        })
Y
Yu Yang 已提交
343 344 345
    return tmp


P
peizhilin 已提交
346
if os.name != 'nt':
P
peizhilin 已提交
347

P
peizhilin 已提交
348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434
    @templatedoc(op_type="lstm")
    def dynamic_lstm(input,
                     size,
                     h_0=None,
                     c_0=None,
                     param_attr=None,
                     bias_attr=None,
                     use_peepholes=True,
                     is_reverse=False,
                     gate_activation='sigmoid',
                     cell_activation='tanh',
                     candidate_activation='tanh',
                     dtype='float32',
                     name=None):
        """
        ${comment}

        Args:
            input (Variable): ${input_comment}
            size (int): 4 * hidden size.
            h_0(Variable): The initial hidden state is an optional input, default is zero.
                           This is a tensor with shape (N x D), where N is the
                           batch size and D is the hidden size.
            c_0(Variable): The initial cell state is an optional input, default is zero.
                           This is a tensor with shape (N x D), where N is the
                           batch size. `h_0` and `c_0` can be NULL but only at the same time.
            param_attr(ParamAttr|None): The parameter attribute for the learnable
                                   hidden-hidden weights.

                                   - Weights = {:math:`W_{ch}, W_{ih}, \
                                                    W_{fh}, W_{oh}`}
                                   - The shape is (D x 4D), where D is the hidden
                                     size.

                                   If it is set to None or one attribute of ParamAttr,
                                   dynamic_lstm will create ParamAttr as param_attr.
                                   If the Initializer of the param_attr is not set, the
                                   parameter is initialized with Xavier. Default: None.
            bias_attr (ParamAttr|None): The bias attribute for the learnable bias
                                  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`}.
                                     - The shape is (1 x 4D).
                                  2. `use_peepholes = True`
                                     - Biases = { :math:`b_c, b_i, b_f, b_o, W_{ic}, \
                                                     W_{fc}, W_{oc}`}.
                                     - The shape is (1 x 7D).

                                  If it is set to None or one attribute of ParamAttr,
                                  dynamic_lstm will create ParamAttr as bias_attr.
                                  If the Initializer of the bias_attr is not set,
                                  the bias is initialized zero. Default: None.
            use_peepholes (bool): ${use_peepholes_comment}
            is_reverse (bool): ${is_reverse_comment}
            gate_activation (str): ${gate_activation_comment}
            cell_activation (str): ${cell_activation_comment}
            candidate_activation (str): ${candidate_activation_comment}
            dtype (str): Data type. Choices = ["float32", "float64"], default "float32".
            name (str|None): A name for this layer(optional). If set None, the layer
                             will be named automatically.

        Returns:
            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`.

        Examples:
            .. code-block:: python

                hidden_dim = 512
                forward_proj = fluid.layers.fc(input=input_seq, size=hidden_dim * 4,
                                               bias_attr=False)
                forward, _ = fluid.layers.dynamic_lstm(
                    input=forward_proj, size=hidden_dim * 4, use_peepholes=False)
        """
        assert bias_attr is not False, "bias_attr should not be False in dynamic_lstmp."
        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)
Y
Yu Yang 已提交
435

P
peizhilin 已提交
436 437 438 439 440 441 442 443 444 445 446 447 448 449
        hidden = helper.create_variable_for_type_inference(dtype)
        cell = helper.create_variable_for_type_inference(dtype)
        batch_gate = helper.create_variable_for_type_inference(dtype)
        batch_cell_pre_act = helper.create_variable_for_type_inference(dtype)
        inputs = {'Input': input, 'Weight': weight, 'Bias': bias}
        batch_size = input.shape[0]
        if h_0:
            assert h_0.shape == (batch_size, size), \
                'The shape of h0 should be (batch_size, %d)' % size
            inputs['H0'] = h_0
        if c_0:
            assert c_0.shape == (batch_size, size), \
                'The shape of c0 should be (batch_size, %d)' % size
            inputs['C0'] = c_0
Y
Yu Yang 已提交
450

P
peizhilin 已提交
451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467
        helper.append_op(
            type='lstm',
            inputs=inputs,
            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
Yu Yang 已提交
468 469


Y
Yibing Liu 已提交
470 471 472 473 474 475 476 477 478 479 480
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',
481 482
                  dtype='float32',
                  name=None):
Y
Yibing Liu 已提交
483 484 485
    """
    **Dynamic LSTMP Layer**

486 487 488 489 490 491
    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
492 493 494 495 496

    The formula is as follows:

    .. math::

497
        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
498

499
        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
500

501
        \\tilde{c_t} & = act_g(W_{cx}x_t + W_{cr}r_{t-1} + b_c)
Y
Yibing Liu 已提交
502

503
        o_t & = \sigma(W_{ox}x_{t} + W_{or}r_{t-1} + W_{oc}c_t + b_o)
Y
Yibing Liu 已提交
504

505
        c_t & = f_t \odot c_{t-1} + i_t \odot \\tilde{c_t}
Y
Yibing Liu 已提交
506

507
        h_t & = o_t \odot act_h(c_t)
Y
Yibing Liu 已提交
508

509
        r_t & = \overline{act_h}(W_{rh}h_t)
Y
Yibing Liu 已提交
510

Y
Yibing Liu 已提交
511 512 513 514 515 516
    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, \
517
          we use vectors to reprenset these diagonal weight matrices.
Y
Yibing Liu 已提交
518
    * :math:`b`: Denotes bias vectors (e.g. :math:`b_i` is the input gate \
519
          bias vector).
Y
Yibing Liu 已提交
520 521 522
    * :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 \
523
          the same size as the cell output activation vector :math:`h`.
Y
Yibing Liu 已提交
524
    * :math:`h`: The hidden state.
525
    * :math:`r`: The recurrent projection of the hidden state.
Y
Yibing Liu 已提交
526 527
    * :math:`\\tilde{c_t}`: The candidate hidden state, whose \
          computation is based on the current input and previous hidden state.
528
    * :math:`\odot`: The element-wise product of the vectors.
Y
Yibing Liu 已提交
529
    * :math:`act_g` and :math:`act_h`: The cell input and cell output \
530
          activation functions and `tanh` is usually used for them.
Y
Yibing Liu 已提交
531 532
    * :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
533 534 535 536

    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.
537

Y
Yibing Liu 已提交
538 539 540 541 542 543 544 545 546 547 548 549
    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.
550
        param_attr(ParamAttr|None): The parameter attribute for the learnable
Y
Yibing Liu 已提交
551 552
                               hidden-hidden weight and projection weight.

553 554
                               - Hidden-hidden weight = {:math:`W_{ch}, W_{ih}, \
                                                W_{fh}, W_{oh}`}.
555 556
                               - 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
557 558
                                 size.
                               - Projection weight = {:math:`W_{rh}`}.
559
                               - The shape of projection weight is (D x P).
C
chengduo 已提交
560 561 562 563 564

                               If it is set to None or one attribute of ParamAttr,
                               dynamic_lstm will create ParamAttr as param_attr.
                               If the Initializer of the param_attr is not set, the
                               parameter is initialized with Xavier. Default: None.
565
        bias_attr(ParamAttr|None): The bias attribute for the learnable bias
Y
Yibing Liu 已提交
566 567 568 569 570 571
                              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`}.
572
                                - The shape is (1 x 4D).
Y
Yibing Liu 已提交
573 574 575
                              2. `use_peepholes = True`
                                - Biases = { :math:`b_c, b_i, b_f, b_o, W_{ic}, \
                                                 W_{fc}, W_{oc}`}.
576
                                - The shape is (1 x 7D).
C
chengduo 已提交
577 578 579 580 581

                              If it is set to None or one attribute of ParamAttr,
                              dynamic_lstm will create ParamAttr as bias_attr.
                              If the Initializer of the bias_attr is not set,
                              the bias is initialized zero. Default: None.
Y
Yibing Liu 已提交
582 583 584 585 586 587 588 589 590
        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.
591
                              Choices = ["sigmoid", "tanh", "relu", "identity"],
Y
Yibing Liu 已提交
592 593
                              default "tanh".
        proj_activation(str): The activation for projection output.
594
                              Choices = ["sigmoid", "tanh", "relu", "identity"],
Y
Yibing Liu 已提交
595 596
                              default "tanh".
        dtype(str): Data type. Choices = ["float32", "float64"], default "float32".
597 598
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.
Y
Yibing Liu 已提交
599 600

    Returns:
601 602 603 604
        tuple: A tuple of two output variable: the projection of hidden state, \
               and cell state of LSTMP. The shape of projection is (T x P), \
               for the cell state which is (T x D), and both LoD is the same \
               with the `input`.
Y
Yibing Liu 已提交
605 606

    Examples:
607

Y
Yibing Liu 已提交
608 609
        .. code-block:: python

610 611 612 613
            dict_dim, emb_dim = 128, 64
            data = fluid.layers.data(name='sequence', shape=[1],
                                     dtype='int32', lod_level=1)
            emb = fluid.layers.embedding(input=data, size=[dict_dim, emb_dim])
Y
Yibing Liu 已提交
614
            hidden_dim, proj_dim = 512, 256
615
            fc_out = fluid.layers.fc(input=emb, size=hidden_dim * 4,
Y
Yibing Liu 已提交
616
                                     act=None, bias_attr=None)
617 618 619
            proj_out, _ = fluid.layers.dynamic_lstmp(input=fc_out,
                                                     size=hidden_dim * 4,
                                                     proj_size=proj_dim,
Y
Yibing Liu 已提交
620 621 622 623
                                                     use_peepholes=False,
                                                     is_reverse=True,
                                                     cell_activation="tanh",
                                                     proj_activation="tanh")
Y
Yibing Liu 已提交
624
    """
625

C
chengduo 已提交
626
    assert bias_attr is not False, "bias_attr should not be False in dynamic_lstmp."
Y
Yibing Liu 已提交
627
    helper = LayerHelper('lstmp', **locals())
M
minqiyang 已提交
628
    size = size // 4
Y
Yibing Liu 已提交
629 630 631 632 633 634 635 636 637 638
    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)

X
Xin Pan 已提交
639 640 641 642 643 644
    projection = helper.create_variable_for_type_inference(dtype)
    cell = helper.create_variable_for_type_inference(dtype)
    ordered_proj0 = helper.create_variable_for_type_inference(dtype)
    batch_hidden = helper.create_variable_for_type_inference(dtype)
    batch_gate = helper.create_variable_for_type_inference(dtype)
    batch_cell_pre_act = helper.create_variable_for_type_inference(dtype)
Y
Yibing Liu 已提交
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

    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
673 674 675 676 677 678 679 680 681
def dynamic_gru(input,
                size,
                param_attr=None,
                bias_attr=None,
                is_reverse=False,
                gate_activation='sigmoid',
                candidate_activation='tanh',
                h_0=None):
    """
682
    **Gated Recurrent Unit (GRU) Layer**
G
guosheng 已提交
683

684
    Refer to `Empirical Evaluation of Gated Recurrent Neural Networks on
685
    Sequence Modeling <https://arxiv.org/abs/1412.3555>`_ .
686

G
guosheng 已提交
687 688 689 690 691 692 693 694 695
    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)
696

G
guosheng 已提交
697
        h_t & = (1-u_t) \odot h_{t-1} + u_t \odot \\tilde{h_t}
698

G
guosheng 已提交
699
    The :math:`\odot` is the element-wise product of the vectors. :math:`act_g`
700 701
    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
702 703 704 705
    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
706
    to use fully-connect layer before GRU layer.
G
guosheng 已提交
707 708

    Args:
709 710
        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
711
            Variable is a matrix with shape :math:`(T \\times 3D)`, where
712
            :math:`T` is the total time steps in this mini-batch, :math:`D`
G
guosheng 已提交
713 714
            is the hidden size.
        size(int): The dimension of the gru cell.
715
        param_attr(ParamAttr|None): The parameter attribute for the learnable
G
guosheng 已提交
716 717
            hidden-hidden weight matrix. Note:

718
            - The shape of the weight matrix is :math:`(T \\times 3D)`, where
G
guosheng 已提交
719
              :math:`D` is the hidden size.
720
            - All elements in the weight matrix can be divided into two parts.
G
guosheng 已提交
721
              The first part are weights of the update gate and reset gate with
722
              shape :math:`(D \\times 2D)`, and the second part are weights for
G
guosheng 已提交
723
              candidate hidden state with shape :math:`(D \\times D)`.
724 725 726 727 728 729 730 731 732 733 734 735

            If it is set to None or one attribute of ParamAttr, dynamic_gru will
            create ParamAttr as param_attr. If the Initializer of the param_attr
            is not set, the parameter is initialized with Xavier. Default: None.
        bias_attr (ParamAttr|bool|None): The parameter attribute for the bias
            of GRU. Note that the bias with :math:`(1 \\times 3D)` concatenates 
            the bias in the update gate, reset gate and candidate calculations.
            If it is set to False, no bias will be applied to the update gate, 
            reset gate and candidate calculations. If it is set to None or one 
            attribute of ParamAttr, dynamic_gru will create ParamAttr as 
            bias_attr. If the Initializer of the bias_attr is not set, the bias
            is initialized zero. Default: None.
736
        is_reverse(bool): Whether to compute reversed GRU, default
G
guosheng 已提交
737 738 739
            :attr:`False`.
        gate_activation(str): The activation for update gate and reset gate.
            Choices = ["sigmoid", "tanh", "relu", "identity"], default "sigmoid".
740
        candidate_activation(str): The activation for candidate hidden state.
G
guosheng 已提交
741
            Choices = ["sigmoid", "tanh", "relu", "identity"], default "tanh".
742 743 744 745
        h_0 (Variable): This is initial hidden state. If not set, default is
            zero. This is a tensor with shape (N x D), where N is the number of
            total time steps of input mini-batch feature and D is the hidden
            size.
G
guosheng 已提交
746 747

    Returns:
G
guosheng 已提交
748
        Variable: The hidden state of GRU. The shape is :math:`(T \\times D)`, \
749
            and sequence length is the same with the input.
750

G
guosheng 已提交
751
    Examples:
752

G
guosheng 已提交
753 754
        .. code-block:: python

755 756 757 758
            dict_dim, emb_dim = 128, 64
            data = fluid.layers.data(name='sequence', shape=[1],
                                     dtype='int32', lod_level=1)
            emb = fluid.layers.embedding(input=data, size=[dict_dim, emb_dim])
G
guosheng 已提交
759
            hidden_dim = 512
760
            x = fluid.layers.fc(input=emb, size=hidden_dim * 3)
G
guosheng 已提交
761 762 763 764 765 766 767 768 769 770
            hidden = fluid.layers.dynamic_gru(input=x, dim=hidden_dim)
    """

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

    weight = helper.create_parameter(
        attr=helper.param_attr, shape=[size, 3 * size], dtype=dtype)
    bias = helper.create_parameter(
        attr=helper.bias_attr, shape=[1, 3 * size], dtype=dtype, is_bias=True)
Y
Yancey 已提交
771
    batch_size = input.shape[0]
G
guosheng 已提交
772
    inputs = {'Input': input, 'Weight': weight, 'Bias': bias}
S
sneaxiy 已提交
773
    if h_0:
G
guosheng 已提交
774
        assert h_0.shape == (
Y
Yancey 已提交
775 776 777
            batch_size, size
        ), 'The shape of h0 should be(batch_size, %d)' % size
        inputs['H0'] = h_0
G
guosheng 已提交
778

X
Xin Pan 已提交
779 780 781 782
    hidden = helper.create_variable_for_type_inference(dtype)
    batch_gate = helper.create_variable_for_type_inference(dtype)
    batch_reset_hidden_prev = helper.create_variable_for_type_inference(dtype)
    batch_hidden = helper.create_variable_for_type_inference(dtype)
G
guosheng 已提交
783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800

    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
801 802 803
def gru_unit(input,
             hidden,
             size,
804 805
             param_attr=None,
             bias_attr=None,
Y
Yu Yang 已提交
806
             activation='tanh',
807
             gate_activation='sigmoid'):
Y
Yu Yang 已提交
808
    """
809
    GRU unit layer. The equation of a gru step is:
Y
Yu Yang 已提交
810

811 812
        .. math::
            u_t & = actGate(xu_{t} + W_u h_{t-1} + b_u)
Y
Yu Yang 已提交
813

814
            r_t & = actGate(xr_{t} + W_r h_{t-1} + b_r)
Y
Yu Yang 已提交
815

816
            m_t & = actNode(xm_t + W_c dot(r_t, h_{t-1}) + b_m)
817

818
            h_t & = dot((1-u_t), m_t) + dot(u_t, h_{t-1})
819 820

    The inputs of gru unit includes :math:`z_t`, :math:`h_{t-1}`. In terms
821 822 823
    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
824 825
    connected layer has to be applied, such that :math:`z_t = W_{fc}x_t`.

826 827
    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
828 829 830
    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`.
831 832 833

    Args:
        input (Variable): The fc transformed input value of current step.
834
        hidden (Variable): The hidden value of gru unit from previous step.
835
        size (integer): The input dimension value.
836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856
        param_attr(ParamAttr|None): The parameter attribute for the learnable
            hidden-hidden weight matrix. Note:

            - The shape of the weight matrix is :math:`(T \\times 3D)`, where
              :math:`D` is the hidden size.
            - All elements in the weight matrix can be divided into two parts.
              The first part are weights of the update gate and reset gate with
              shape :math:`(D \\times 2D)`, and the second part are weights for
              candidate hidden state with shape :math:`(D \\times D)`.

            If it is set to None or one attribute of ParamAttr, gru_unit will
            create ParamAttr as param_attr. If the Initializer of the param_attr
            is not set, the parameter is initialized with Xavier. Default: None.
        bias_attr (ParamAttr|bool|None): The parameter attribute for the bias
            of GRU. Note that the bias with :math:`(1 \\times 3D)` concatenates 
            the bias in the update gate, reset gate and candidate calculations.
            If it is set to False, no bias will be applied to the update gate, 
            reset gate and candidate calculations. If it is set to None or one 
            attribute of ParamAttr, gru_unit will create ParamAttr as 
            bias_attr. If the Initializer of the bias_attr is not set, the bias
            is initialized zero. Default: None.
857 858 859 860
        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
861

862 863 864 865 866 867
    Returns:
        tuple: The hidden value, reset-hidden value and gate values.

    Examples:

        .. code-block:: python
Y
Yu Yang 已提交
868

869
             # assuming we have x_t_data and prev_hidden of size=10
870
             x_t = fluid.layers.fc(input=x_t_data, size=30)
871 872
             hidden_val, r_h_val, gate_val = fluid.layers.gru_unit(input=x_t,
                                                    hidden = prev_hidden)
Y
Yu Yang 已提交
873 874 875 876 877 878 879 880 881 882 883 884

    """
    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()
M
minqiyang 已提交
885
    size = size // 3
Y
Yu Yang 已提交
886 887

    # create weight
888 889
    weight = helper.create_parameter(
        attr=helper.param_attr, shape=[size, 3 * size], dtype=dtype)
Y
Yu Yang 已提交
890

X
Xin Pan 已提交
891 892 893
    gate = helper.create_variable_for_type_inference(dtype)
    reset_hidden_pre = helper.create_variable_for_type_inference(dtype)
    updated_hidden = helper.create_variable_for_type_inference(dtype)
894
    inputs = {'Input': input, 'HiddenPrev': hidden, 'Weight': weight}
Y
Yu Yang 已提交
895
    # create bias
896
    if helper.bias_attr:
Y
Yu Yang 已提交
897 898 899
        bias_size = [1, 3 * size]
        bias = helper.create_parameter(
            attr=helper.bias_attr, shape=bias_size, dtype=dtype, is_bias=True)
900
        inputs['Bias'] = bias
Y
Yu Yang 已提交
901 902 903

    helper.append_op(
        type='gru_unit',
904
        inputs=inputs,
Y
Yu Yang 已提交
905 906 907 908 909 910
        outputs={
            'Gate': gate,
            'ResetHiddenPrev': reset_hidden_pre,
            'Hidden': updated_hidden,
        },
        attrs={
911 912
            'activation': 2,  # tanh
            'gate_activation': 1,  # sigmoid
Y
Yu Yang 已提交
913 914 915 916 917
        })

    return updated_hidden, reset_hidden_pre, gate


Y
yuyang18 已提交
918
@templatedoc()
919
def linear_chain_crf(input, label, param_attr=None):
Y
yuyang18 已提交
920 921 922 923 924 925 926
    """
    Linear Chain CRF.

    ${comment}

    Args:
        input(${emission_type}): ${emission_comment}
D
dzhwinter 已提交
927
        input(${transition_type}): ${transition_comment}
Y
yuyang18 已提交
928 929 930 931
        label(${label_type}): ${label_comment}
        param_attr(ParamAttr): The attribute of the learnable parameter.

    Returns:
D
dzhwinter 已提交
932 933 934
        output(${emission_exps_type}): ${emission_exps_comment} \n
        output(${transition_exps_type}): ${transition_exps_comment} \n
        output(${log_likelihood_type}): ${log_likelihood_comment}
Y
yuyang18 已提交
935 936

    """
Y
Yu Yang 已提交
937 938 939 940 941 942
    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())
X
Xin Pan 已提交
943 944 945 946 947 948 949 950
    alpha = helper.create_variable_for_type_inference(
        dtype=helper.input_dtype())
    emission_exps = helper.create_variable_for_type_inference(
        dtype=helper.input_dtype())
    transition_exps = helper.create_variable_for_type_inference(
        dtype=helper.input_dtype())
    log_likelihood = helper.create_variable_for_type_inference(
        dtype=helper.input_dtype())
Y
Yu Yang 已提交
951 952 953 954 955 956 957 958 959 960 961 962 963 964 965
    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


P
peizhilin 已提交
966
if os.name != 'nt':
P
peizhilin 已提交
967

P
peizhilin 已提交
968 969 970 971
    @templatedoc()
    def crf_decoding(input, param_attr, label=None):
        """
        ${comment}
Y
yuyang18 已提交
972

P
peizhilin 已提交
973 974
        Args:
            input(${emission_type}): ${emission_comment}
Y
yi.wu 已提交
975

P
peizhilin 已提交
976
            param_attr(ParamAttr): The parameter attribute for training.
Y
yi.wu 已提交
977

P
peizhilin 已提交
978
            label(${label_type}): ${label_comment}
Y
yuyang18 已提交
979

P
peizhilin 已提交
980 981
        Returns:
            Variable: ${viterbi_path_comment}
982

P
peizhilin 已提交
983 984
        Examples:
            .. code-block:: python
Y
yi.wu 已提交
985

P
peizhilin 已提交
986 987 988 989 990 991 992 993 994
               crf_decode = layers.crf_decoding(
                    input=hidden, param_attr=ParamAttr(name="crfw"))
        """
        helper = LayerHelper('crf_decoding', **locals())
        transition = helper.get_parameter(param_attr.name)
        viterbi_path = helper.create_variable_for_type_inference(
            dtype=helper.input_dtype())
        helper.append_op(
            type='crf_decoding',
P
peizhilin 已提交
995 996 997 998 999
            inputs={
                "Emission": [input],
                "Transition": transition,
                "Label": label
            },
P
peizhilin 已提交
1000
            outputs={"ViterbiPath": [viterbi_path]})
Y
Yu Yang 已提交
1001

P
peizhilin 已提交
1002
        return viterbi_path
Y
Yu Yang 已提交
1003 1004


Y
yi.wu 已提交
1005
@templatedoc()
F
fengjiayi 已提交
1006
def cos_sim(X, Y):
Y
Yu Yang 已提交
1007
    """
Y
yi.wu 已提交
1008 1009 1010
    ${comment}

    Args:
1011 1012
        X (Variable): ${x_comment}.
        Y (Variable): ${y_comment}.
F
fengjiayi 已提交
1013

Y
yi.wu 已提交
1014
    Returns:
1015
        Variable: the output of cosine(X, Y).
Y
Yu Yang 已提交
1016
    """
F
fengjiayi 已提交
1017
    helper = LayerHelper('cos_sim', **locals())
X
Xin Pan 已提交
1018 1019 1020
    out = helper.create_variable_for_type_inference(dtype=X.dtype)
    xnorm = helper.create_variable_for_type_inference(dtype=X.dtype)
    ynorm = helper.create_variable_for_type_inference(dtype=X.dtype)
Y
Yu Yang 已提交
1021 1022 1023 1024 1025 1026 1027 1028 1029 1030
    helper.append_op(
        type='cos_sim',
        inputs={'X': [X],
                'Y': [Y]},
        outputs={'Out': [out],
                 'XNorm': [xnorm],
                 'YNorm': [ynorm]})
    return out


P
phlrain 已提交
1031 1032 1033 1034 1035
def dropout(x,
            dropout_prob,
            is_test=False,
            seed=None,
            name=None,
P
phlrain 已提交
1036
            dropout_implementation="downgrade_in_infer"):
1037 1038 1039 1040 1041
    """
    Computes dropout.

    Drop or keep each element of `x` independently. Dropout is a regularization
    technique for reducing overfitting by preventing neuron co-adaption during
1042
    training. The dropout operator randomly sets (according to the given dropout
1043 1044 1045 1046
    probability) the outputs of some units to zero, while others are remain
    unchanged.

    Args:
1047 1048
        x (Variable): The input tensor variable.
        dropout_prob (float): Probability of setting units to zero.
1049 1050 1051 1052 1053 1054 1055
        is_test (bool): A flag indicating whether it is in test phrase or not.
        seed (int): A Python integer used to create random seeds. If this
                    parameter is set to None, a random seed is used.
                    NOTE: If an integer seed is given, always the same output
                    units will be dropped. DO NOT use a fixed seed in training.
        name (str|None): A name for this layer(optional). If set None, the layer
                         will be named automatically.
P
phlrain 已提交
1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069
        dropout_implementation(string): ['downgrade_in_infer'(defauld)|'upscale_in_train']
                                        1. downgrade_in_infer(default), downgrade the outcome at inference
                                           train: out = input * mask
                                           inference: out = input * dropout_prob
                                           (make is a tensor same shape with input, value is 0 or 1
                                            ratio of 0 is dropout_prob)
                                        2. upscale_in_train, upscale the outcome at training time
                                           train: out = input * mask / ( 1.0 - dropout_prob )
                                           inference: out = input
                                           (make is a tensor same shape with input, value is 0 or 1
                                            ratio of 0 is dropout_prob)
                                           dropout op can be removed from the program. 
                                           the program will be efficient
                                        
P
phlrain 已提交
1070

1071 1072

    Returns:
1073
        Variable: A tensor variable is the shape with `x`.
1074 1075

    Examples:
1076

1077 1078
        .. code-block:: python

1079 1080
            x = fluid.layers.data(name="data", shape=[32, 32], dtype="float32")
            droped = fluid.layers.dropout(x, dropout_prob=0.5)
1081 1082
    """

F
fengjiayi 已提交
1083
    helper = LayerHelper('dropout', **locals())
X
Xin Pan 已提交
1084 1085 1086
    out = helper.create_variable_for_type_inference(dtype=x.dtype)
    mask = helper.create_variable_for_type_inference(
        dtype=x.dtype, stop_gradient=True)
C
chengduo 已提交
1087 1088 1089 1090

    if (seed is None or seed == 0) and helper.main_program.random_seed != 0:
        seed = helper.main_program.random_seed

1091 1092 1093 1094 1095
    helper.append_op(
        type='dropout',
        inputs={'X': [x]},
        outputs={'Out': [out],
                 'Mask': [mask]},
1096 1097 1098 1099
        attrs={
            'dropout_prob': dropout_prob,
            'is_test': is_test,
            'fix_seed': seed is not None,
P
phlrain 已提交
1100 1101
            'seed': seed if seed is not None else 0,
            'dropout_implementation': dropout_implementation,
1102
        })
1103 1104 1105
    return out


1106
def cross_entropy(input, label, soft_label=False, ignore_index=-100):
Y
Yu Yang 已提交
1107
    """
Y
Yibing Liu 已提交
1108 1109
    **Cross Entropy Layer**

1110 1111 1112
    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
1113 1114

    1) One-hot cross-entropy:
F
fengjiayi 已提交
1115
        `soft_label = False`, `Label[i, 0]` indicates the class index for sample i:
Y
yangyaming 已提交
1116

Y
Yibing Liu 已提交
1117
        .. math::
Y
yangyaming 已提交
1118

Y
Yibing Liu 已提交
1119 1120 1121
            Y[i] = -\log(X[i, Label[i]])

    2) Soft-label cross-entropy:
F
fengjiayi 已提交
1122 1123
        `soft_label = True`, `Label[i, j]` indicates the soft label of class j
        for sample i:
Y
Yibing Liu 已提交
1124 1125 1126 1127 1128

        .. math::

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

Y
Yibing Liu 已提交
1129
       Please make sure that in this case the summation of each row of `label`
Y
Yibing Liu 已提交
1130 1131 1132
       equals one.

    3) One-hot cross-entropy with vecterized `label`:
F
fengjiayi 已提交
1133 1134
         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
1135
         to a one-hot cross-entropy with one-hot label representation.
Y
yangyaming 已提交
1136

Y
Yibing Liu 已提交
1137
    Args:
Y
yangyaming 已提交
1138
        input (Variable|list):  a 2-D tensor with shape [N x D], where N is the
1139 1140 1141 1142
                                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
1143
        label (Variable|list): the ground truth which is a 2-D tensor. When
1144 1145 1146 1147
                               `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
1148
        soft_label (bool): a flag indicating whether to
1149
                                           interpretate the given labels as soft
1150
                                           labels. Default: `False`.
M
minqiyang 已提交
1151 1152
        ignore_index (int): Specifies a target value that is ignored and does
                            not contribute to the input gradient. Only valid
1153
                            if soft_label is set to False. Default: -100
Y
Yibing Liu 已提交
1154 1155 1156 1157 1158

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

    Raises:
1159 1160 1161 1162 1163
        `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
1164 1165 1166 1167 1168 1169

    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
1170
    """
F
fengjiayi 已提交
1171
    helper = LayerHelper('cross_entropy', **locals())
X
Xin Pan 已提交
1172
    out = helper.create_variable_for_type_inference(dtype=input.dtype)
Y
Yu Yang 已提交
1173 1174 1175 1176 1177
    helper.append_op(
        type='cross_entropy',
        inputs={'X': [input],
                'Label': [label]},
        outputs={'Y': [out]},
1178 1179
        attrs={"soft_label": soft_label,
               "ignore_index": ignore_index})
Y
Yu Yang 已提交
1180 1181 1182
    return out


F
fengjiayi 已提交
1183
def square_error_cost(input, label):
Y
Yu Yang 已提交
1184
    """
1185 1186
    **Square error cost layer**

1187 1188
    This layer accepts input predictions and target label and returns the
    squared error cost.
Y
ying 已提交
1189

1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202
    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:
1203 1204
        input (Variable): Input tensor, has predictions.
        label (Variable): Label tensor, has target labels.
1205 1206

    Returns:
G
guosheng 已提交
1207
        Variable: The tensor variable storing the element-wise squared error \
1208
                  difference of input and label.
1209 1210 1211 1212 1213 1214 1215 1216

    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
1217
    """
F
fengjiayi 已提交
1218
    helper = LayerHelper('square_error_cost', **locals())
X
Xin Pan 已提交
1219
    minus_out = helper.create_variable_for_type_inference(dtype=input.dtype)
Y
Yu Yang 已提交
1220 1221 1222 1223 1224 1225
    helper.append_op(
        type='elementwise_sub',
        inputs={'X': [input],
                'Y': [label]},
        outputs={'Out': [minus_out]})

X
Xin Pan 已提交
1226
    square_out = helper.create_variable_for_type_inference(dtype=input.dtype)
Y
Yu Yang 已提交
1227
    helper.append_op(
F
fengjiayi 已提交
1228 1229
        type='square', inputs={'X': [minus_out]},
        outputs={'Out': [square_out]})
Y
Yu Yang 已提交
1230 1231 1232
    return square_out


Y
yi.wu 已提交
1233
@templatedoc()
Y
Yu Yang 已提交
1234 1235 1236 1237
def chunk_eval(input,
               label,
               chunk_scheme,
               num_chunk_types,
F
fengjiayi 已提交
1238
               excluded_chunk_types=None):
Y
Yu Yang 已提交
1239
    """
Y
yi.wu 已提交
1240
    **Chunk Evaluator**
Y
yi.wu 已提交
1241

Y
yangyaming 已提交
1242
    This function computes and outputs the precision, recall and
1243
    F1-score of chunk detection.
Y
yi.wu 已提交
1244

Y
yi.wu 已提交
1245 1246 1247 1248 1249 1250 1251 1252
    For some basics of chunking, please refer to
    'Chunking with Support Vector Machines <https://aclanthology.info/pdf/N/N01/N01-1025.pdf>'.

    ChunkEvalOp computes the precision, recall, and F1-score of chunk detection,
    and supports IOB, IOE, IOBES and IO (also known as plain) tagging schemes.
    Here is a NER example of labeling for these tagging schemes:

    .. code-block:: python
1253

Y
yi.wu 已提交
1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278
       ====== ====== ======  =====  ==  ============   =====  ===== =====  ==  =========
              Li     Ming    works  at  Agricultural   Bank   of    China  in  Beijing.
       ====== ====== ======  =====  ==  ============   =====  ===== =====  ==  =========
       IO     I-PER  I-PER   O      O   I-ORG          I-ORG  I-ORG I-ORG  O   I-LOC
       IOB    B-PER  I-PER   O      O   B-ORG          I-ORG  I-ORG I-ORG  O   B-LOC
       IOE    I-PER  E-PER   O      O   I-ORG          I-ORG  I-ORG E-ORG  O   E-LOC
       IOBES  B-PER  E-PER   O      O   I-ORG          I-ORG  I-ORG E-ORG  O   S-LOC
       ====== ====== ======  =====  ==  ============   =====  ===== =====  ==  =========

    There are three chunk types(named entity types) including PER(person), ORG(organization)
    and LOC(LOCATION), and we can see that the labels have the form <tag type>-<chunk type>.

    Since the calculations actually use label ids rather than labels, extra attention
    should be paid when mapping labels to ids to make CheckEvalOp work. The key point
    is that the listed equations are satisfied by ids.

    .. code-block:: python

       tag_type = label % num_tag_type
       chunk_type = label / num_tag_type

    where `num_tag_type` is the num of tag types in the tagging scheme, `num_chunk_type`
    is the num of chunk types, and `tag_type` get its value from the following table.

    .. code-block:: python
1279

Y
yi.wu 已提交
1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303
       Scheme Begin Inside End   Single
        plain   0     -      -     -
        IOB     0     1      -     -
        IOE     -     0      1     -
        IOBES   0     1      2     3

    Still use NER as example, assuming the tagging scheme is IOB while chunk types are ORG,
    PER and LOC. To satisfy the above equations, the label map can be like this:

    .. code-block:: python

       B-ORG  0
       I-ORG  1
       B-PER  2
       I-PER  3
       B-LOC  4
       I-LOC  5
       O      6

    It's not hard to verify the equations noting that the num of chunk types
    is 3 and the num of tag types in IOB scheme is 2. For example, the label
    id of I-LOC is 5, the tag type id of I-LOC is 1, and the chunk type id of
    I-LOC is 2, which consistent with the results from the equations.

Y
yi.wu 已提交
1304
    Args:
1305 1306 1307 1308 1309
        input (Variable): prediction output of the network.
        label (Variable): label of the test data set.
        chunk_scheme (str): ${chunk_scheme_comment}
        num_chunk_types (int): ${num_chunk_types_comment}
        excluded_chunk_types (list): ${excluded_chunk_types_comment}
F
fengjiayi 已提交
1310

Y
yi.wu 已提交
1311
    Returns:
Y
update  
yi.wu 已提交
1312 1313 1314
        tuple: tuple containing: precision, recall, f1_score,
        num_infer_chunks, num_label_chunks,
        num_correct_chunks
1315

Y
yi.wu 已提交
1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327
    Examples:
        .. code-block:: python

            crf = fluid.layers.linear_chain_crf(
                input=hidden, label=label, param_attr=ParamAttr(name="crfw"))
            crf_decode = fluid.layers.crf_decoding(
                input=hidden, param_attr=ParamAttr(name="crfw"))
            fluid.layers.chunk_eval(
                input=crf_decode,
                label=label,
                chunk_scheme="IOB",
                num_chunk_types=(label_dict_len - 1) / 2)
Y
Yu Yang 已提交
1328
    """
F
fengjiayi 已提交
1329
    helper = LayerHelper("chunk_eval", **locals())
Y
Yu Yang 已提交
1330 1331

    # prepare output
X
Xin Pan 已提交
1332 1333 1334 1335 1336 1337 1338
    precision = helper.create_variable_for_type_inference(dtype="float32")
    recall = helper.create_variable_for_type_inference(dtype="float32")
    f1_score = helper.create_variable_for_type_inference(dtype="float32")
    num_infer_chunks = helper.create_variable_for_type_inference(dtype="int64")
    num_label_chunks = helper.create_variable_for_type_inference(dtype="int64")
    num_correct_chunks = helper.create_variable_for_type_inference(
        dtype="int64")
Y
Yu Yang 已提交
1339 1340 1341 1342 1343 1344 1345 1346

    helper.append_op(
        type="chunk_eval",
        inputs={"Inference": [input],
                "Label": [label]},
        outputs={
            "Precision": [precision],
            "Recall": [recall],
1347 1348 1349 1350
            "F1-Score": [f1_score],
            "NumInferChunks": [num_infer_chunks],
            "NumLabelChunks": [num_label_chunks],
            "NumCorrectChunks": [num_correct_chunks]
Y
Yu Yang 已提交
1351 1352 1353
        },
        attrs={
            "num_chunk_types": num_chunk_types,
G
guosheng 已提交
1354 1355
            "chunk_scheme": chunk_scheme,
            "excluded_chunk_types": excluded_chunk_types or []
Y
Yu Yang 已提交
1356
        })
1357 1358
    return (precision, recall, f1_score, num_infer_chunks, num_label_chunks,
            num_correct_chunks)
Y
Yu Yang 已提交
1359 1360


1361
@templatedoc()
Y
Yu Yang 已提交
1362 1363 1364 1365 1366 1367 1368
def sequence_conv(input,
                  num_filters,
                  filter_size=3,
                  filter_stride=1,
                  padding=None,
                  bias_attr=None,
                  param_attr=None,
C
chengduo 已提交
1369 1370
                  act=None,
                  name=None):
Y
Yu Yang 已提交
1371 1372 1373 1374
    """
    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.
1375 1376 1377 1378 1379 1380 1381

    Args:
        input (Variable): ${x_comment}
        num_filters (int): number of filters.
        filter_size (int): the filter size (H and W).
        filter_stride (int): stride of the filter.
        padding (bool): if True, add paddings.
C
chengduo 已提交
1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394
        bias_attr (ParamAttr|bool|None): The parameter attribute for the bias of sequence_conv.
            If it is set to False, no bias will be added to the output units.
            If it is set to None or one attribute of ParamAttr, sequence_conv
            will create ParamAttr as bias_attr. If the Initializer of the bias_attr
            is not set, the bias is initialized zero. Default: None.
        param_attr (ParamAttr|None): The parameter attribute for learnable parameters/weights
            of sequence_conv. If it is set to None or one attribute of ParamAttr, sequence_conv
            will create ParamAttr as param_attr. If the Initializer of the param_attr
            is not set, the parameter is initialized with Xavier. Default: None.
        act (str): Activation type, if it is set to None, activation is not appended.
            Default: None.
        name (str|None): A name for this layer(optional). If set None, the layer
            will be named automatically. Default: None.
F
fengjiayi 已提交
1395

1396 1397
    Returns:
        Variable: output of sequence_conv
Y
Yu Yang 已提交
1398 1399 1400 1401 1402 1403 1404
    """

    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)
X
Xin Pan 已提交
1405
    pre_bias = helper.create_variable_for_type_inference(dtype)
Y
Yu Yang 已提交
1406 1407 1408 1409 1410 1411 1412 1413 1414 1415

    helper.append_op(
        type='sequence_conv',
        inputs={
            'X': [input],
            'Filter': [filter_param],
        },
        outputs={"Out": pre_bias},
        attrs={
            'contextStride': filter_stride,
M
minqiyang 已提交
1416
            'contextStart': -int(filter_size // 2),
Y
Yu Yang 已提交
1417 1418 1419 1420 1421 1422
            'contextLength': filter_size
        })
    pre_act = helper.append_bias_op(pre_bias)
    return helper.append_activation(pre_act)


C
chengduo 已提交
1423
def sequence_softmax(input, use_cudnn=False, name=None):
1424 1425 1426
    """
    This function computes the softmax activation among all time-steps for each
    sequence. The dimension of each time-step should be 1. Thus, the shape of
1427
    input Tensor can be either :math:`[N, 1]` or :math:`[N]`, where :math:`N`
1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443
    is the sum of the length of all sequences.

    For i-th sequence in a mini-batch:

    .. math::

        Out(X[lod[i]:lod[i+1]], :) = \\frac{\exp(X[lod[i]:lod[i+1], :])}{\sum(\exp(X[lod[i]:lod[i+1], :]))}

    For example, for a mini-batch of 3 sequences with variable-length,
    each containing 2, 3, 2 time-steps, the lod of which is [0, 2, 5, 7],
    then softmax will be computed among :math:`X[0:2, :]`, :math:`X[2:5, :]`,
    :math:`X[5:7, :]`, and :math:`N` turns out to be 7.

    Args:
        input (Variable): The input variable which is a LoDTensor.
        use_cudnn (bool): Use cudnn kernel or not, it is valid only when the cudnn \
C
chengduo 已提交
1444 1445 1446
            library is installed. Default: False.
        name (str|None): A name for this layer(optional). If set None, the layer
            will be named automatically. Default: None.
1447

1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458
    Returns:
        Variable: output of sequence_softmax

    Examples:

        .. code-block:: python

             x = fluid.layers.data(name='x', shape=[7, 1],
                              dtype='float32', lod_level=1)
             x_sequence_softmax = fluid.layers.sequence_softmax(input=x)
    """
1459 1460
    helper = LayerHelper('sequence_softmax', **locals())
    dtype = helper.input_dtype()
X
Xin Pan 已提交
1461
    softmax_out = helper.create_variable_for_type_inference(dtype)
1462 1463 1464 1465 1466 1467 1468 1469
    helper.append_op(
        type="sequence_softmax",
        inputs={"X": input},
        outputs={"Out": softmax_out},
        attrs={"use_cudnn": use_cudnn})
    return softmax_out


C
chengduo 已提交
1470
def softmax(input, use_cudnn=True, name=None):
Q
qiaolongfei 已提交
1471
    """
1472
    The input of the softmax operator is a tensor of any rank. The output tensor
F
fengjiayi 已提交
1473
    has the same shape as the input.
Q
qiaolongfei 已提交
1474

1475 1476 1477 1478 1479 1480
    The input tensor will first be logically flattened to a 2-D matrix. The matrix's
    second dimension(row length) is as same as the last dimension of the input
    tensor, and the first dimension(column length) is the product of all other
    dimensions of the input tensor. For each row of the matrix, the softmax operator
    squashes the K-dimensional(K is the width of the matrix, which is also the size
    of the input tensor's last dimension) vector of arbitrary real values to a
F
fengjiayi 已提交
1481
    K-dimensional vector of real values in the range [0, 1] that add up to 1.
Q
qiaolongfei 已提交
1482 1483 1484 1485 1486 1487 1488

    It computes the exponential of the given dimension and the sum of exponential
    values of all the other dimensions in the K-dimensional vector input.
    Then the ratio of the exponential of the given dimension and the sum of
    exponential values of all the other dimensions is the output of the softmax
    operator.

F
fengjiayi 已提交
1489
    For each row :math:`i` and each column :math:`j` in the matrix, we have:
Q
qiaolongfei 已提交
1490 1491 1492 1493 1494 1495 1496 1497

    .. math::

        Out[i, j] = \\frac{\exp(X[i, j])}{\sum_j(exp(X[i, j])}

    Args:
        input (Variable): The input variable.
        use_cudnn (bool): Use cudnn kernel or not, it is valid only when the cudnn \
C
chengduo 已提交
1498 1499 1500
            library is installed.
        name (str|None): A name for this layer(optional). If set None, the layer
            will be named automatically. Default: None.
Q
qiaolongfei 已提交
1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512

    Returns:
        Variable: output of softmax

    Examples:

        .. code-block:: python

             fc = fluid.layers.fc(input=x, size=10)
             softmax = fluid.layers.softmax(input=fc)

    """
1513 1514
    helper = LayerHelper('softmax', **locals())
    dtype = helper.input_dtype()
X
Xin Pan 已提交
1515
    softmax_out = helper.create_variable_for_type_inference(dtype)
1516 1517 1518 1519 1520 1521 1522 1523
    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
1524 1525 1526
def conv2d(input,
           num_filters,
           filter_size,
C
chengduoZH 已提交
1527 1528
           stride=1,
           padding=0,
1529
           dilation=1,
Y
Yu Yang 已提交
1530 1531 1532
           groups=None,
           param_attr=None,
           bias_attr=None,
C
chengduoZH 已提交
1533
           use_cudnn=True,
1534 1535
           act=None,
           name=None):
Y
Yu Yang 已提交
1536
    """
C
chengduoZH 已提交
1537
    The convolution2D layer calculates the output based on the input, filter
T
tensor-tang 已提交
1538 1539
    and strides, paddings, dilations, groups parameters. Input and
    Output are in NCHW format, where N is batch size, C is the number of
1540
    channels, H is the height of the feature, and W is the width of the feature.
T
tensor-tang 已提交
1541 1542 1543 1544 1545 1546 1547
    Filter is in MCHW format, where M is the number of output image channels,
    C is the number of input image channels, H is the height of the filter,
    and W is the width of the filter. If the groups is greater than 1,
    C will equal the number of input image channels divided by the groups.
    Please refer to UFLDL's `convolution
    <http://ufldl.stanford.edu/tutorial/supervised/FeatureExtractionUsingConvolution/>`_
    for more detials.
1548 1549 1550
    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
1551

1552
    For each input :math:`X`, the equation is:
C
refine  
chengduoZH 已提交
1553

C
chengduoZH 已提交
1554 1555
    .. math::

C
refine  
chengduoZH 已提交
1556
        Out = \sigma (W \\ast X + b)
C
chengduoZH 已提交
1557

T
tensor-tang 已提交
1558
    Where:
C
chengduoZH 已提交
1559

1560 1561 1562 1563 1564
    * :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.
T
tensor-tang 已提交
1565
    * :math:`Out`: Output value, the shape of :math:`Out` and :math:`X` may be different.
C
chengduoZH 已提交
1566 1567 1568

    Example:

1569 1570
        - Input:

W
weixing02 已提交
1571
          Input shape: :math:`(N, C_{in}, H_{in}, W_{in})`
C
refine  
chengduoZH 已提交
1572

W
weixing02 已提交
1573
          Filter shape: :math:`(C_{out}, C_{in}, H_f, W_f)`
C
refine  
chengduoZH 已提交
1574

1575
        - Output:
T
tensor-tang 已提交
1576

W
weixing02 已提交
1577
          Output shape: :math:`(N, C_{out}, H_{out}, W_{out})`
C
refine  
chengduoZH 已提交
1578

C
chengduoZH 已提交
1579
        Where
1580 1581

        .. math::
C
chengduoZH 已提交
1582

W
weixing02 已提交
1583 1584
            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
1585 1586

    Args:
1587
        input (Variable): The input image with [N, C, H, W] format.
T
tensor-tang 已提交
1588
        num_filters(int): The number of filter. It is as same as the output
1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605
            image channel.
        filter_size (int|tuple|None): The filter size. If filter_size is a tuple,
            it must contain two integers, (filter_size_H, filter_size_W).
            Otherwise, the filter will be a square.
        stride (int|tuple): The stride size. If stride is a tuple, it must
            contain two integers, (stride_H, stride_W). Otherwise, the
            stride_H = stride_W = stride. Default: stride = 1.
        padding (int|tuple): The padding size. If padding is a tuple, it must
            contain two integers, (padding_H, padding_W). Otherwise, the
            padding_H = padding_W = padding. Default: padding = 0.
        dilation (int|tuple): The dilation size. If dilation is a tuple, it must
            contain two integers, (dilation_H, dilation_W). Otherwise, the
            dilation_H = dilation_W = dilation. Default: dilation = 1.
        groups (int): The groups number of the Conv2d Layer. According to grouped
            convolution in Alex Krizhevsky's Deep CNN paper: when group=2,
            the first half of the filters is only connected to the first half
            of the input channels, while the second half of the filters is only
C
chengduo 已提交
1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616
            connected to the second half of the input channels. Default: groups=1.
        param_attr (ParamAttr|None): The parameter attribute for learnable parameters/weights
            of conv2d. If it is set to None or one attribute of ParamAttr, conv2d
            will create ParamAttr as param_attr. If the Initializer of the param_attr
            is not set, the parameter is initialized with :math:`Normal(0.0, std)`,
             and the :math:`std` is :math:`(\\frac{2.0 }{filter\_elem\_num})^{0.5}`. Default: None.
        bias_attr (ParamAttr|bool|None): The parameter attribute for the bias of conv2d.
            If it is set to False, no bias will be added to the output units.
            If it is set to None or one attribute of ParamAttr, conv2d
            will create ParamAttr as bias_attr. If the Initializer of the bias_attr
            is not set, the bias is initialized zero. Default: None.
1617 1618
        use_cudnn (bool): Use cudnn kernel or not, it is valid only when the cudnn
            library is installed. Default: True
C
chengduo 已提交
1619 1620
        act (str): Activation type, if it is set to None, activation is not appended.
            Default: None
1621
        name (str|None): A name for this layer(optional). If set None, the layer
C
chengduo 已提交
1622
            will be named automatically. Default: None
C
chengduoZH 已提交
1623 1624

    Returns:
G
guosheng 已提交
1625
        Variable: The tensor variable storing the convolution and \
C
chengduoZH 已提交
1626 1627
                  non-linearity activation result.

C
refine  
chengduoZH 已提交
1628
    Raises:
1629 1630
        ValueError: If the shapes of input, filter_size, stride, padding and
                    groups mismatch.
C
refine  
chengduoZH 已提交
1631

C
chengduoZH 已提交
1632 1633 1634
    Examples:
        .. code-block:: python

1635 1636
          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
1637 1638 1639
    """

    num_channels = input.shape[1]
C
chengduo 已提交
1640
    assert param_attr is not False, "param_attr should not be False here."
1641
    l_type = 'conv2d'
X
xzl 已提交
1642 1643
    if (num_channels == groups and num_filters % num_channels == 0 and
            not use_cudnn):
1644
        l_type = 'depthwise_conv2d'
1645 1646 1647 1648

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

Y
Yu Yang 已提交
1649 1650 1651 1652 1653
    if groups is None:
        num_filter_channels = num_channels
    else:
        if num_channels % groups != 0:
            raise ValueError("num_channels must be divisible by groups.")
M
minqiyang 已提交
1654
        num_filter_channels = num_channels // groups
Y
Yu Yang 已提交
1655

C
chengduoZH 已提交
1656 1657 1658
    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')
1659
    dilation = utils.convert_to_list(dilation, 2, 'dilation')
C
chengduoZH 已提交
1660

C
chengduoZH 已提交
1661 1662
    if not isinstance(use_cudnn, bool):
        raise ValueError("use_cudnn should be True or False")
Y
Yu Yang 已提交
1663 1664

    input_shape = input.shape
M
minqiyang 已提交
1665
    filter_shape = [num_filters, int(num_filter_channels)] + filter_size
Y
Yu Yang 已提交
1666 1667

    def _get_default_param_initializer():
C
chengduo 已提交
1668 1669
        filter_elem_num = filter_size[0] * filter_size[1] * num_channels
        std = (2.0 / filter_elem_num)**0.5
Y
Yu Yang 已提交
1670 1671 1672 1673 1674 1675 1676 1677
        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())

X
Xin Pan 已提交
1678
    pre_bias = helper.create_variable_for_type_inference(dtype)
Y
Yu Yang 已提交
1679

1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693
    if use_cudnn:
        helper.create_variable(
            name="kCUDNNFwdAlgoCache",
            persistable=True,
            type=core.VarDesc.VarType.RAW)
        helper.create_variable(
            name="kCUDNNBwdDataAlgoCache",
            persistable=True,
            type=core.VarDesc.VarType.RAW)
        helper.create_variable(
            name="kCUDNNBwdFilterAlgoCache",
            persistable=True,
            type=core.VarDesc.VarType.RAW)

Y
Yu Yang 已提交
1694
    helper.append_op(
1695
        type=l_type,
Y
Yu Yang 已提交
1696 1697 1698 1699 1700
        inputs={
            'Input': input,
            'Filter': filter_param,
        },
        outputs={"Output": pre_bias},
C
chengduoZH 已提交
1701 1702 1703
        attrs={
            'strides': stride,
            'paddings': padding,
1704
            'dilations': dilation,
C
chengduoZH 已提交
1705
            'groups': groups,
1706
            'use_cudnn': use_cudnn,
1707
            'use_mkldnn': False,
C
chengduoZH 已提交
1708
        })
Y
Yu Yang 已提交
1709 1710 1711 1712 1713 1714

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

    return helper.append_activation(pre_act)


C
chengduoZH 已提交
1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731
def conv3d(input,
           num_filters,
           filter_size,
           stride=1,
           padding=0,
           dilation=1,
           groups=None,
           param_attr=None,
           bias_attr=None,
           use_cudnn=True,
           act=None,
           name=None):
    """
    **Convlution3D Layer**

    The convolution3D layer calculates the output based on the input, filter
    and strides, paddings, dilations, groups parameters. Input(Input) and
1732 1733 1734 1735 1736 1737
    Output(Output) are in NCDHW format. Where N is batch size C is the number of
    channels, D is the depth of the feature, H is the height of the feature,
    and W is the width of the feature. Convlution3D is similar with Convlution2D
    but adds one dimension(depth). If bias attribution and activation type are
    provided, bias is added to the output of the convolution, and the
    corresponding activation function is applied to the final result.
C
chengduoZH 已提交
1738 1739 1740 1741 1742 1743 1744 1745 1746

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

    .. math::

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

    In the above equation:

1747 1748
    * :math:`X`: Input value, a tensor with NCDHW format.
    * :math:`W`: Filter value, a tensor with MCDHW format.
C
chengduoZH 已提交
1749 1750 1751
    * :math:`\\ast`: Convolution operation.
    * :math:`b`: Bias value, a 2-D tensor with shape [M, 1].
    * :math:`\\sigma`: Activation function.
1752
    * :math:`Out`: Output value, the shape of :math:`Out` and :math:`X` may be different.
C
chengduoZH 已提交
1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777

    Example:

        - Input:

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

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

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

        Where

        .. math::

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

    Args:
        input (Variable): The input image with [N, C, D, H, W] format.
            num_filters(int): The number of filter. It is as same as the output
            image channel.
        filter_size (int|tuple|None): The filter size. If filter_size is a tuple,
1778
            it must contain three integers, (filter_size_D, filter_size_H, filter_size_W).
C
chengduoZH 已提交
1779 1780
            Otherwise, the filter will be a square.
        stride (int|tuple): The stride size. If stride is a tuple, it must
1781
            contain three integers, (stride_D, stride_H, stride_W). Otherwise, the
C
chengduoZH 已提交
1782 1783
            stride_D = stride_H = stride_W = stride. Default: stride = 1.
        padding (int|tuple): The padding size. If padding is a tuple, it must
1784
            contain three integers, (padding_D, padding_H, padding_W). Otherwise, the
C
chengduoZH 已提交
1785 1786
            padding_D = padding_H = padding_W = padding. Default: padding = 0.
        dilation (int|tuple): The dilation size. If dilation is a tuple, it must
1787
            contain three integers, (dilation_D, dilation_H, dilation_W). Otherwise, the
C
chengduoZH 已提交
1788 1789 1790 1791 1792 1793
            dilation_D = dilation_H = dilation_W = dilation. Default: dilation = 1.
        groups (int): The groups number of the Conv3d Layer. According to grouped
            convolution in Alex Krizhevsky's Deep CNN paper: when group=2,
            the first half of the filters is only connected to the first half
            of the input channels, while the second half of the filters is only
            connected to the second half of the input channels. Default: groups=1
C
chengduo 已提交
1794 1795 1796 1797 1798 1799 1800 1801 1802 1803
        param_attr (ParamAttr|None): The parameter attribute for learnable parameters/weights
            of conv3d. If it is set to None or one attribute of ParamAttr, conv3d
            will create ParamAttr as param_attr. If it is set to None, the parameter
            is initialized with :math:`Normal(0.0, std)`, and the :math:`std` is
            :math:`(\\frac{2.0 }{filter\_elem\_num})^{0.5}`. Default: None.
        bias_attr (ParamAttr|bool|None): The parameter attribute for the bias of conv3d.
            If it is set to False, no bias will be added to the output units.
            If it is set to None or one attribute of ParamAttr, conv3d
            will create ParamAttr as bias_attr. If the Initializer of the bias_attr
            is not set, the bias is initialized zero. Default: None.
C
chengduoZH 已提交
1804 1805
        use_cudnn (bool): Use cudnn kernel or not, it is valid only when the cudnn
            library is installed. Default: True
C
chengduo 已提交
1806 1807
        act (str): Activation type, if it is set to None, activation is not appended.
            Default: None.
C
chengduoZH 已提交
1808
        name (str|None): A name for this layer(optional). If set None, the layer
C
chengduo 已提交
1809
            will be named automatically. Default: None.
C
chengduoZH 已提交
1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821

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

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

    Examples:
        .. code-block:: python

1822 1823
          data = fluid.layers.data(name='data', shape=[3, 12, 32, 32], dtype='float32')
          conv3d = fluid.layers.conv3d(input=data, num_filters=2, filter_size=3, act="relu")
C
chengduoZH 已提交
1824 1825 1826
    """

    l_type = 'conv3d'
C
chengduo 已提交
1827
    assert param_attr is not False, "param_attr should not be False here."
C
chengduoZH 已提交
1828 1829 1830 1831 1832 1833 1834 1835 1836 1837
    helper = LayerHelper(l_type, **locals())
    dtype = helper.input_dtype()

    num_channels = input.shape[1]

    if groups is None:
        num_filter_channels = num_channels
    else:
        if num_channels % groups != 0:
            raise ValueError("num_channels must be divisible by groups.")
M
minqiyang 已提交
1838
        num_filter_channels = num_channels // groups
C
chengduoZH 已提交
1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851

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

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

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

    def _get_default_param_initializer():
C
chengduo 已提交
1852 1853 1854
        filter_elem_num = filter_size[0] * filter_size[1] * filter_size[
            2] * num_channels
        std = (2.0 / filter_elem_num)**0.5
C
chengduoZH 已提交
1855 1856 1857 1858 1859 1860 1861 1862
        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())

X
Xin Pan 已提交
1863
    pre_bias = helper.create_variable_for_type_inference(dtype)
C
chengduoZH 已提交
1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877

    helper.append_op(
        type=l_type,
        inputs={
            'Input': input,
            'Filter': filter_param,
        },
        outputs={"Output": pre_bias},
        attrs={
            'strides': stride,
            'paddings': padding,
            'dilations': dilation,
            'groups': groups,
            'use_cudnn': use_cudnn,
X
Xin Pan 已提交
1878
            'use_mkldnn': False
C
chengduoZH 已提交
1879 1880
        })

1881
    pre_act = helper.append_bias_op(pre_bias, dim_start=1, dim_end=2)
C
chengduoZH 已提交
1882 1883 1884 1885

    return helper.append_activation(pre_act)


J
Jacek Czaja 已提交
1886
def sequence_pool(input, pool_type, is_test=False):
Y
Yu Yang 已提交
1887
    """
Y
yangyaming 已提交
1888 1889 1890
    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
1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901

    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:
1902
         x.lod = [[2, 3, 2]]
L
Luo Tao 已提交
1903 1904 1905 1906 1907
         x.data = [1, 3, 2, 4, 6, 5, 1]
         x.dims = [7, 1]

       then output is a Tensor:
         out.dim = [3, 1]
1908
         with condition len(x.lod[-1]) == out.dims[0]
L
Luo Tao 已提交
1909 1910 1911 1912 1913 1914 1915

       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)
1916 1917
         last   : out.data = [3, 6, 1], where 3=last(1,3), 6=last(2,4,6), 1=last(5,1)
         first  : out.data = [1, 2, 5], where 1=first(1,3), 2=first(2,4,6), 5=first(5,1)
F
fengjiayi 已提交
1918

L
Luo Tao 已提交
1919 1920
    Args:
        input(variable): The input variable which is a LoDTensor.
Y
yangyaming 已提交
1921
        pool_type (string): The pooling type of sequence_pool.
L
Luo Tao 已提交
1922
            It supports average, sum, sqrt and max.
J
Jacek Czaja 已提交
1923
        is_test(bool, Default False): Used distinguish training from scoring mode.
L
Luo Tao 已提交
1924 1925 1926 1927 1928 1929 1930

    Returns:
        The sequence pooling variable which is a Tensor.

    Examples:

        .. code-block:: python
F
fengjiayi 已提交
1931

Y
yangyaming 已提交
1932
             x = fluid.layers.data(name='x', shape=[7, 1],
L
Luo Tao 已提交
1933 1934 1935 1936 1937
                              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')
1938 1939
             last_x = fluid.layers.sequence_pool(input=x, pool_type='last')
             first_x = fluid.layers.sequence_pool(input=x, pool_type='first')
Y
Yu Yang 已提交
1940
    """
F
fengjiayi 已提交
1941
    helper = LayerHelper('sequence_pool', **locals())
Y
Yu Yang 已提交
1942
    dtype = helper.input_dtype()
X
Xin Pan 已提交
1943 1944
    pool_out = helper.create_variable_for_type_inference(dtype)
    max_index = helper.create_variable_for_type_inference(dtype)
Y
Yu Yang 已提交
1945 1946 1947 1948 1949 1950

    helper.append_op(
        type="sequence_pool",
        inputs={"X": input},
        outputs={"Out": pool_out,
                 "MaxIndex": max_index},
J
Jacek Czaja 已提交
1951 1952
        attrs={"pooltype": pool_type.upper(),
               "is_test": is_test})
Y
Yu Yang 已提交
1953

Y
yangyaming 已提交
1954 1955 1956 1957 1958
    # 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
1959 1960 1961
    return pool_out


C
add doc  
chengduoZH 已提交
1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980
@templatedoc()
def sequence_concat(input, name=None):
    """
    ${comment}

    Args:
        input(list): List of Variables to be concatenated.
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.

    Returns:
        Variable: Output variable of the concatenation.

    Examples:
        .. code-block:: python

           out = fluid.layers.sequence_concat(input=[seq1, seq2, seq3])
    """
    helper = LayerHelper('sequence_concat', **locals())
X
Xin Pan 已提交
1981
    out = helper.create_variable_for_type_inference(dtype=helper.input_dtype())
C
add doc  
chengduoZH 已提交
1982 1983 1984 1985 1986
    helper.append_op(
        type='sequence_concat', inputs={'X': input}, outputs={'Out': [out]})
    return out


F
fengjiayi 已提交
1987
def sequence_first_step(input):
L
Luo Tao 已提交
1988
    """
L
Luo Tao 已提交
1989
    This function gets the first step of sequence.
L
Luo Tao 已提交
1990 1991 1992 1993

    .. code-block:: text

       x is a 1-level LoDTensor:
1994
         x.lod = [[2, 3, 2]]
L
Luo Tao 已提交
1995 1996 1997 1998 1999
         x.data = [1, 3, 2, 4, 6, 5, 1]
         x.dims = [7, 1]

       then output is a Tensor:
         out.dim = [3, 1]
2000
         with condition len(x.lod[-1]) == out.dims[0]
L
Luo Tao 已提交
2001
         out.data = [1, 2, 5], where 1=first(1,3), 2=first(2,4,6), 5=first(5,1)
F
fengjiayi 已提交
2002

L
Luo Tao 已提交
2003 2004 2005 2006 2007 2008 2009 2010 2011
    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
2012

Y
yangyaming 已提交
2013
             x = fluid.layers.data(name='x', shape=[7, 1],
L
Luo Tao 已提交
2014 2015 2016
                              dtype='float32', lod_level=1)
             x_first_step = fluid.layers.sequence_first_step(input=x)
    """
2017 2018 2019
    return sequence_pool(input=input, pool_type="first")


F
fengjiayi 已提交
2020
def sequence_last_step(input):
L
Luo Tao 已提交
2021
    """
L
Luo Tao 已提交
2022
    This function gets the last step of sequence.
L
Luo Tao 已提交
2023 2024 2025 2026

    .. code-block:: text

       x is a 1-level LoDTensor:
2027
         x.lod = [[2, 3, 2]]
L
Luo Tao 已提交
2028 2029 2030 2031 2032
         x.data = [1, 3, 2, 4, 6, 5, 1]
         x.dims = [7, 1]

       then output is a Tensor:
         out.dim = [3, 1]
2033
         with condition len(x.lod[-1]) == out.dims[0]
L
Luo Tao 已提交
2034
         out.data = [3, 6, 1], where 3=last(1,3), 6=last(2,4,6), 1=last(5,1)
F
fengjiayi 已提交
2035

L
Luo Tao 已提交
2036 2037 2038 2039 2040 2041 2042 2043 2044
    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
2045

Y
yangyaming 已提交
2046
             x = fluid.layers.data(name='x', shape=[7, 1],
L
Luo Tao 已提交
2047 2048 2049
                              dtype='float32', lod_level=1)
             x_last_step = fluid.layers.sequence_last_step(input=x)
    """
2050 2051 2052
    return sequence_pool(input=input, pool_type="last")


Y
Yibing Liu 已提交
2053 2054 2055 2056
def sequence_slice(input, offset, length, name=None):
    """
    **Sequence Slice Layer**

2057
    The layer crops a subsequence from given sequence with given start
Y
Yibing Liu 已提交
2058 2059 2060 2061 2062
    offset and subsequence length.

    It only supports sequence data (LoDTensor with lod_level equal to 1).

    .. code-block:: text
2063

Y
Yibing Liu 已提交
2064 2065
	- Case:

2066
            Given the input Variable **input**:
2067

2068 2069 2070
                input.data = [[a1, a2], [b1, b2], [c1, c2], [d1, d2], [e1, e2]],
                input.lod = [[3, 2]],
                input.dims = (5, 2),
Y
Yibing Liu 已提交
2071

2072
            with offset.data = [[0], [1]] and length.data = [[2], [1]],
Y
Yibing Liu 已提交
2073

2074
            the output Variable will be
2075

2076 2077 2078
                out.data = [[a1, a2], [b1, b2], [e1, e2]],
                out.lod = [[2, 1]],
                out.dims = (3, 2).
2079 2080

    NOTE: The first dimension size of **input**, **offset** and **length**
2081
          should be equal. The **offset** should start from 0.
2082

Y
Yibing Liu 已提交
2083
    Args:
2084
        input(Variable): The input Variable which consists of the complete
Y
Yibing Liu 已提交
2085
                         sequences.
Y
Yibing Liu 已提交
2086 2087 2088 2089 2090 2091
        offset(Variable): The offset to slice each sequence.
        length(Variable): The length of each subsequence.
        name(str|None): A name for this layer(optional). If set None, the
                        layer will be named automatically.

    Returns:
Y
Yibing Liu 已提交
2092
        Variable: The output subsequences.
Y
Yibing Liu 已提交
2093 2094 2095 2096 2097 2098 2099 2100 2101 2102

    Examples:

        .. code-block:: python

             import numpy as np
             seqs = fluid.layers.data(name='x', shape=[10, 5],
                              dtype='float32', lod_level=1)
             offset = fluid.layers.assign(input=np.array([[0, 1]]).astype("int32"))
             length = fluid.layers.assign(input=np.array([[2, 1]]).astype("int32"))
2103
             subseqs = fluid.layers.sequence_slice(input=seqs, offset=offset,
Y
Yibing Liu 已提交
2104 2105 2106 2107
                                                   length=length)
    """
    helper = LayerHelper("sequence_slice", **locals())
    dtype = helper.input_dtype()
X
Xin Pan 已提交
2108
    out = helper.create_variable_for_type_inference(dtype)
Y
Yibing Liu 已提交
2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122

    offset.stop_gradient = True
    length.stop_gradient = True

    helper.append_op(
        type="sequence_slice",
        inputs={"X": input,
                "Offset": offset,
                "Length": length},
        outputs={"Out": out})

    return out


F
fengjiayi 已提交
2123
@templatedoc()
Y
Yu Yang 已提交
2124
def pool2d(input,
C
chengduoZH 已提交
2125 2126
           pool_size=-1,
           pool_type="max",
C
chengduoZH 已提交
2127 2128
           pool_stride=1,
           pool_padding=0,
C
caoying03 已提交
2129
           global_pooling=False,
C
chengduoZH 已提交
2130
           use_cudnn=True,
2131
           ceil_mode=False,
2132 2133
           name=None,
           exclusive=True):
Y
Yu Yang 已提交
2134
    """
F
fengjiayi 已提交
2135
    ${comment}
2136 2137

    Args:
2138 2139 2140
        input (Variable): The input tensor of pooling operator. The format of
                          input tensor is NCHW, where N is batch size, C is
                          the number of channels, H is the height of the
F
fengjiayi 已提交
2141
                          feature, and W is the width of the feature.
2142
        pool_size (int): The side length of pooling windows. All pooling
F
fengjiayi 已提交
2143
                         windows are squares with pool_size on a side.
F
fengjiayi 已提交
2144
        pool_type: ${pooling_type_comment}
2145 2146
        pool_stride (int): stride of the pooling layer.
        pool_padding (int): padding size.
2147 2148 2149
        global_pooling (bool): ${global_pooling_comment}
        use_cudnn (bool): ${use_cudnn_comment}
        ceil_mode (bool): ${ceil_mode_comment}
2150
        name (str|None): A name for this layer(optional). If set None, the
F
fengjiayi 已提交
2151
                        layer will be named automatically.
2152 2153
        exclusive (bool): Whether to exclude padding points in average pooling 
                          mode, default is true
F
fengjiayi 已提交
2154

2155
    Returns:
F
fengjiayi 已提交
2156
        Variable: The pooling result.
F
fengjiayi 已提交
2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169

    Raises:
        ValueError: If 'pool_type' is not "max" nor "avg"
        ValueError: If 'global_pooling' is False and 'pool_size' is -1
        ValueError: If 'use_cudnn' is not a bool value.

    Examples:

        .. code-block:: python

          data = fluid.layers.data(
              name='data', shape=[3, 32, 32], dtype='float32')
          conv2d = fluid.layers.pool2d(
2170 2171 2172 2173
                            input=data,
                            pool_size=2,
                            pool_type='max',
                            pool_stride=1,
F
fengjiayi 已提交
2174
                            global_pooling=False)
Y
Yu Yang 已提交
2175 2176 2177 2178 2179
    """
    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
2180

C
chengduoZH 已提交
2181 2182 2183 2184 2185
    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
2186 2187 2188 2189
    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
2190 2191
    if not isinstance(use_cudnn, bool):
        raise ValueError("use_cudnn should be True or False")
Y
Yu Yang 已提交
2192

C
Add doc  
chengduoZH 已提交
2193
    l_type = 'pool2d'
2194 2195

    helper = LayerHelper(l_type, **locals())
Y
Yu Yang 已提交
2196
    dtype = helper.input_dtype()
X
Xin Pan 已提交
2197
    pool_out = helper.create_variable_for_type_inference(dtype)
Y
Yu Yang 已提交
2198 2199

    helper.append_op(
2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210
        type=l_type,
        inputs={"X": input},
        outputs={"Out": pool_out},
        attrs={
            "pooling_type": pool_type,
            "ksize": pool_size,
            "global_pooling": global_pooling,
            "strides": pool_stride,
            "paddings": pool_padding,
            "use_cudnn": use_cudnn,
            "ceil_mode": ceil_mode,
2211 2212
            "use_mkldnn": False,
            "exclusive": exclusive,
2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225
        })

    return pool_out


def pool3d(input,
           pool_size=-1,
           pool_type="max",
           pool_stride=1,
           pool_padding=0,
           global_pooling=False,
           use_cudnn=True,
           ceil_mode=False,
2226 2227
           name=None,
           exclusive=True):
2228 2229
    """
    This function adds the operator for pooling in 3-dimensions, using the
Y
Yu Yang 已提交
2230
    pooling configurations mentioned in input parameters.
2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242

    Args:
        input (Variable): ${input_comment}
        pool_size (int): ${ksize_comment}
        pool_type (str): ${pooling_type_comment}
        pool_stride (int): stride of the pooling layer.
        pool_padding (int): padding size.
        global_pooling (bool): ${global_pooling_comment}
        use_cudnn (bool): ${use_cudnn_comment}
        ceil_mode (bool): ${ceil_mode_comment}
        name (str): A name for this layer(optional). If set None, the layer
            will be named automatically.
2243 2244
        exclusive (bool): Whether to exclude padding points in average pooling 
                          mode, default is true
2245

2246
    Returns:
2247
        Variable: output of pool3d layer.
Y
Yu Yang 已提交
2248 2249 2250 2251 2252
    """
    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
2253

C
chengduoZH 已提交
2254 2255 2256 2257 2258
    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))

2259 2260 2261
    pool_size = utils.convert_to_list(pool_size, 3, 'pool_size')
    pool_padding = utils.convert_to_list(pool_padding, 3, 'pool_padding')
    pool_stride = utils.convert_to_list(pool_stride, 3, 'pool_stride')
C
chengduoZH 已提交
2262

C
chengduoZH 已提交
2263 2264
    if not isinstance(use_cudnn, bool):
        raise ValueError("use_cudnn should be True or False")
Y
Yu Yang 已提交
2265

2266 2267
    l_type = "pool3d"
    helper = LayerHelper(l_type, **locals())
Y
Yu Yang 已提交
2268
    dtype = helper.input_dtype()
X
Xin Pan 已提交
2269
    pool_out = helper.create_variable_for_type_inference(dtype)
Y
Yu Yang 已提交
2270 2271

    helper.append_op(
2272
        type=l_type,
Y
Yu Yang 已提交
2273 2274 2275 2276 2277 2278 2279
        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
2280
            "paddings": pool_padding,
2281
            "use_cudnn": use_cudnn,
2282
            "ceil_mode": ceil_mode,
2283 2284
            "use_mkldnn": False,
            "exclusive": exclusive,
Y
Yu Yang 已提交
2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296
        })

    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
2297
               data_layout='NCHW',
Y
Yang Yang 已提交
2298
               in_place=False,
2299 2300
               name=None,
               moving_mean_name=None,
W
wanghaoshuang 已提交
2301
               moving_variance_name=None,
2302 2303
               do_model_average_for_mean_and_var=False,
               fuse_with_relu=False):
Y
Yu Yang 已提交
2304
    """
Q
qiaolongfei 已提交
2305 2306 2307 2308
    **Batch Normalization Layer**

    Can be used as a normalizer function for conv2d and fully_connected operations.
    The required data format for this layer is one of the following:
Q
qiaolongfei 已提交
2309

Q
qiaolongfei 已提交
2310
    1. NHWC `[batch, in_height, in_width, in_channels]`
Q
qiaolongfei 已提交
2311

Q
qiaolongfei 已提交
2312 2313
    2. NCHW `[batch, in_channels, in_height, in_width]`

Q
qiaolongfei 已提交
2314 2315 2316
    Refer to `Batch Normalization: Accelerating Deep Network Training by Reducing
    Internal Covariate Shift <https://arxiv.org/pdf/1502.03167.pdf>`_
    for more details.
Q
qiaolongfei 已提交
2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328

    :math:`input` is the input features over a mini-batch.

    ..  math::

        \\mu_{\\beta} &\\gets \\frac{1}{m} \\sum_{i=1}^{m} x_i \\qquad &//\\
        \ mini-batch\ mean \\\\
        \\sigma_{\\beta}^{2} &\\gets \\frac{1}{m} \\sum_{i=1}^{m}(x_i - \\
        \\mu_{\\beta})^2 \\qquad &//\ mini-batch\ variance \\\\
        \\hat{x_i} &\\gets \\frac{x_i - \\mu_\\beta} {\\sqrt{\\
        \\sigma_{\\beta}^{2} + \\epsilon}} \\qquad &//\ normalize \\\\
        y_i &\\gets \\gamma \\hat{x_i} + \\beta \\qquad &//\ scale\ and\ shift
2329 2330

    Args:
Q
qiaolongfei 已提交
2331
        input(variable): The input variable which is a LoDTensor.
Q
qiaolongfei 已提交
2332 2333 2334 2335
        act(string, Default None): Activation type, linear|relu|prelu|...
        is_test(bool, Default False): Used for training or training.
        momentum(float, Default 0.9):
        epsilon(float, Default 1e-05):
C
chengduo 已提交
2336 2337 2338 2339 2340 2341 2342 2343
        param_attr(ParamAttr|None): The parameter attribute for Parameter `scale`
             of batch_norm. If it is set to None or one attribute of ParamAttr, batch_norm
             will create ParamAttr as param_attr. If the Initializer of the param_attr
             is not set, the parameter is initialized with Xavier. Default: None.
        bias_attr(ParamAttr|None): The parameter attribute for the bias of batch_norm.
             If it is set to None or one attribute of ParamAttr, batch_norm
             will create ParamAttr as bias_attr. If the Initializer of the bias_attr
             is not set, the bias is initialized zero. Default: None.
Q
qiaolongfei 已提交
2344
        data_layout(string, default NCHW): NCHW|NHWC
2345
        in_place(bool, Default False): Make the input and output of batch norm reuse memory.
Q
qiaolongfei 已提交
2346 2347 2348 2349
        name(string, Default None): A name for this layer(optional). If set None, the layer
            will be named automatically.
        moving_mean_name(string, Default None): The name of moving_mean which store the global Mean.
        moving_variance_name(string, Default None): The name of the moving_variance which store the global Variance.
Q
qiaolongfei 已提交
2350
        do_model_average_for_mean_and_var(bool, Default False): Do model average for mean and variance or not.
2351
        fuse_with_relu (bool): if True, this OP performs relu after batch norm.
2352 2353

    Returns:
Q
qiaolongfei 已提交
2354
        Variable: A tensor variable which is the result after applying batch normalization on the input.
Q
qiaolongfei 已提交
2355 2356 2357 2358 2359 2360 2361

    Examples:

        .. code-block:: python

            hidden1 = fluid.layers.fc(input=x, size=200, param_attr='fc1.w')
            hidden2 = fluid.layers.batch_norm(input=hidden1)
Y
Yu Yang 已提交
2362
    """
C
chengduo 已提交
2363
    assert bias_attr is not False, "bias_attr should not be False in batch_norm."
Y
Yu Yang 已提交
2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385
    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(
2386
        attr=helper.bias_attr, shape=param_shape, dtype=dtype, is_bias=True)
Y
Yu Yang 已提交
2387

2388 2389
    mean = helper.create_parameter(
        attr=ParamAttr(
W
wanghaoshuang 已提交
2390 2391 2392
            name=moving_mean_name,
            initializer=Constant(0.0),
            trainable=False,
W
wanghaoshuang 已提交
2393
            do_model_average=do_model_average_for_mean_and_var),
Q
QI JUN 已提交
2394
        shape=param_shape,
2395 2396 2397 2398 2399 2400 2401
        dtype=input.dtype)
    mean.stop_gradient = True

    variance = helper.create_parameter(
        attr=ParamAttr(
            name=moving_variance_name,
            initializer=Constant(1.0),
W
wanghaoshuang 已提交
2402
            trainable=False,
W
wanghaoshuang 已提交
2403
            do_model_average=do_model_average_for_mean_and_var),
Q
QI JUN 已提交
2404
        shape=param_shape,
2405 2406
        dtype=input.dtype)
    variance.stop_gradient = True
Y
Yu Yang 已提交
2407 2408 2409 2410 2411 2412

    # create output
    # mean and mean_out share the same memory
    mean_out = mean
    # variance and variance out share the same memory
    variance_out = variance
X
Xin Pan 已提交
2413 2414 2415 2416
    saved_mean = helper.create_variable_for_type_inference(
        dtype=dtype, stop_gradient=True)
    saved_variance = helper.create_variable_for_type_inference(
        dtype=dtype, stop_gradient=True)
Y
Yu Yang 已提交
2417

X
Xin Pan 已提交
2418 2419
    batch_norm_out = input if in_place else helper.create_variable_for_type_inference(
        dtype)
Y
Yu Yang 已提交
2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436

    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
        },
2437 2438 2439 2440
        attrs={
            "momentum": momentum,
            "epsilon": epsilon,
            "is_test": is_test,
X
Xin Pan 已提交
2441
            "use_mkldnn": False,
2442
            "fuse_with_relu": fuse_with_relu
2443
        })
Y
Yu Yang 已提交
2444 2445 2446 2447

    return helper.append_activation(batch_norm_out)


Y
yuyang18 已提交
2448
@templatedoc()
G
guosheng 已提交
2449 2450 2451 2452 2453 2454 2455 2456 2457 2458
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):
    """
Y
yuyang18 已提交
2459
    ${comment}
G
guosheng 已提交
2460 2461 2462

    The formula is as follows:

Y
yuyang18 已提交
2463
    ..  math::
G
guosheng 已提交
2464 2465 2466 2467 2468 2469 2470

        \\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)

Y
yuyang18 已提交
2471 2472 2473 2474 2475 2476 2477 2478
    * :math:`a`: the vector representation of the summed inputs to the neurons
    in that layer.

    * :math:`H`: the number of hidden units in a layers

    * :math:`g`: the trainable scale parameter.

    * :math:`b`: the trainable bias parameter.
Y
yuyang18 已提交
2479

G
guosheng 已提交
2480 2481
    Args:
        input(Variable): The input tensor variable.
2482
        scale(bool): Whether to learn the adaptive gain :math:`g` after
S
sneaxiy 已提交
2483
            normalization. Default True.
2484
        shift(bool): Whether to learn the adaptive bias :math:`b` after
S
sneaxiy 已提交
2485 2486
            normalization. Default True.
        begin_norm_axis(int): The normalization will be performed along
G
guosheng 已提交
2487
            dimensions from :attr:`begin_norm_axis` to :attr:`rank(input)`.
S
sneaxiy 已提交
2488
            Default 1.
2489
        epsilon(float): The small value added to the variance to prevent
S
sneaxiy 已提交
2490
            division by zero. Default 1e-05.
G
guosheng 已提交
2491
        param_attr(ParamAttr|None): The parameter attribute for the learnable
S
sneaxiy 已提交
2492 2493
            gain :math:`g`. If :attr:`scale` is False, :attr:`param_attr` is
            omitted. If :attr:`scale` is True and :attr:`param_attr` is None,
2494 2495
            a default :code:`ParamAttr` would be added as scale. The
            :attr:`param_attr` is initialized as 1 if it is added. Default None.
G
guosheng 已提交
2496
        bias_attr(ParamAttr|None): The parameter attribute for the learnable
S
sneaxiy 已提交
2497 2498
            bias :math:`b`. If :attr:`shift` is False, :attr:`bias_attr` is
            omitted. If :attr:`shift` is True and :attr:`param_attr` is None,
2499
            a default :code:`ParamAttr` would be added as bias. The
S
sneaxiy 已提交
2500
            :attr:`bias_attr` is initialized as 0 if it is added. Default None.
G
guosheng 已提交
2501
        act(str): Activation to be applied to the output of layer normalizaiton.
S
sneaxiy 已提交
2502 2503 2504
                  Default None.
        name(str): The name of this layer. It is optional. Default None, and a
                   unique name would be generated automatically.
G
guosheng 已提交
2505 2506

    Returns:
Y
yuyang18 已提交
2507
        ${y_comment}
G
guosheng 已提交
2508 2509 2510

    Examples:

Y
yuyang18 已提交
2511 2512 2513
        >>> data = fluid.layers.data(name='data', shape=[3, 32, 32],
        >>>                          dtype='float32')
        >>> x = fluid.layers.layer_norm(input=data, begin_norm_axis=1)
G
guosheng 已提交
2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528
    """
    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
2529
    if shift:
G
guosheng 已提交
2530 2531 2532 2533 2534 2535
        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
X
Xin Pan 已提交
2536 2537 2538 2539 2540
    mean_out = helper.create_variable_for_type_inference(
        dtype=dtype, stop_gradient=True)
    variance_out = helper.create_variable_for_type_inference(
        dtype=dtype, stop_gradient=True)
    layer_norm_out = helper.create_variable_for_type_inference(dtype)
G
guosheng 已提交
2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555

    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)


Y
Yu Yang 已提交
2556 2557 2558 2559
def conv2d_transpose(input,
                     num_filters,
                     output_size=None,
                     filter_size=None,
C
chengduoZH 已提交
2560 2561 2562
                     padding=0,
                     stride=1,
                     dilation=1,
2563
                     groups=None,
C
caoying03 已提交
2564
                     param_attr=None,
2565
                     bias_attr=None,
C
chengduoZH 已提交
2566
                     use_cudnn=True,
2567
                     act=None,
C
caoying03 已提交
2568
                     name=None):
Y
Yu Yang 已提交
2569
    """
2570 2571 2572 2573 2574 2575 2576 2577
    **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
2578 2579
    layer, please refer to the following explanation and references
    `therein <http://www.matthewzeiler.com/wp-content/uploads/2017/07/cvpr2010.pdf>`_.
2580 2581 2582
    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.
2583 2584 2585 2586 2587

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

    .. math::

2588
        Out = \sigma (W \\ast X + b)
2589

2590
    Where:
2591 2592 2593

    * :math:`X`: Input value, a tensor with NCHW format.
    * :math:`W`: Filter value, a tensor with MCHW format.
2594 2595 2596 2597
    * :math:`\\ast`: Convolution operation.
    * :math:`b`: Bias value, a 2-D tensor with shape [M, 1].
    * :math:`\\sigma`: Activation function.
    * :math:`Out`: Output value, the shape of :math:`Out` and :math:`X` may be different.
Y
Yu Yang 已提交
2598

2599 2600 2601 2602
    Example:

        - Input:

2603
          Input shape: :math:`(N, C_{in}, H_{in}, W_{in})`
2604

2605
          Filter shape: :math:`(C_{in}, C_{out}, H_f, W_f)`
2606 2607 2608

        - Output:

2609
          Output shape: :math:`(N, C_{out}, H_{out}, W_{out})`
2610 2611

        Where
Y
Yu Yang 已提交
2612

2613 2614
        .. math::

2615 2616 2617 2618
           H^\prime_{out} &= (H_{in} - 1) * strides[0] - 2 * paddings[0] + dilations[0] * (H_f - 1) + 1 \\\\
           W^\prime_{out} &= (W_{in} - 1) * strides[1] - 2 * paddings[1] + dilations[1] * (W_f - 1) + 1 \\\\
           H_{out} \in [ H^\prime_{out}, H^\prime_{out} + strides[0] ) \\\\
           W_{out} \in [ W^\prime_{out}, W^\prime_{out} + strides[1] )
Y
Yu Yang 已提交
2619 2620

    Args:
2621 2622 2623 2624
        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
2625 2626 2627 2628
            tuple, it must contain two integers, (image_H, image_W). None if use
            filter_size, padding, and stride to calculate output_size.
            if output_size and filter_size are specified at the same time, They
            should follow the formula above.
2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646
        filter_size(int|tuple|None): The filter size. If filter_size is a tuple,
            it must contain two integers, (filter_size_H, filter_size_W).
            Otherwise, the filter will be a square. None if use output size to
            calculate filter_size.
        padding(int|tuple): The padding size. If padding is a tuple, it must
            contain two integers, (padding_H, padding_W). Otherwise, the
            padding_H = padding_W = padding. Default: padding = 0.
        stride(int|tuple): The stride size. If stride is a tuple, it must
            contain two integers, (stride_H, stride_W). Otherwise, the
            stride_H = stride_W = stride. Default: stride = 1.
        dilation(int|tuple): The dilation size. If dilation is a tuple, it must
            contain two integers, (dilation_H, dilation_W). Otherwise, the
            dilation_H = dilation_W = dilation. Default: dilation = 1.
        groups(int): The groups number of the Conv2d transpose layer. Inspired by
            grouped convolution in Alex Krizhevsky's Deep CNN paper, in which
            when group=2, the first half of the filters is only connected to the
            first half of the input channels, while the second half of the
            filters is only connected to the second half of the input channels.
C
chengduo 已提交
2647 2648 2649 2650 2651 2652 2653 2654 2655 2656
            Default: groups = 1.
        param_attr (ParamAttr|None): The parameter attribute for learnable parameters/weights
            of conv2d_transpose. If it is set to None or one attribute of ParamAttr, conv2d_transpose
            will create ParamAttr as param_attr. If the Initializer of the param_attr
            is not set, the parameter is initialized with Xavier. Default: None.
        bias_attr (ParamAttr|bool|None): The parameter attribute for the bias of conv2d_transpose.
            If it is set to False, no bias will be added to the output units.
            If it is set to None or one attribute of ParamAttr, conv2d_transpose
            will create ParamAttr as bias_attr. If the Initializer of the bias_attr
            is not set, the bias is initialized zero. Default: None.
2657
        use_cudnn(bool): Use cudnn kernel or not, it is valid only when the cudnn
C
chengduo 已提交
2658 2659 2660
            library is installed. Default: True.
        act (str): Activation type, if it is set to None, activation is not appended.
            Default: None.
2661
        name(str|None): A name for this layer(optional). If set None, the layer
C
chengduo 已提交
2662
            will be named automatically. Default: True.
Y
Yu Yang 已提交
2663 2664

    Returns:
2665
        Variable: The tensor variable storing the convolution transpose result.
2666 2667

    Raises:
2668 2669
        ValueError: If the shapes of input, filter_size, stride, padding and
                    groups mismatch.
2670 2671 2672 2673

    Examples:
       .. code-block:: python

2674 2675
          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
2676
    """
C
chengduo 已提交
2677
    assert param_attr is not False, "param_attr should not be False in conv2d_transpose."
2678 2679 2680 2681 2682 2683 2684 2685
    input_channel = input.shape[1]

    op_type = 'conv2d_transpose'
    if (input_channel == groups and num_filters == input_channel and
            not use_cudnn):
        op_type = 'depthwise_conv2d_transpose'

    helper = LayerHelper(op_type, **locals())
Y
Yu Yang 已提交
2686 2687 2688
    if not isinstance(input, Variable):
        raise TypeError("Input of conv2d_transpose must be Variable")

C
chengduoZH 已提交
2689 2690 2691
    padding = utils.convert_to_list(padding, 2, 'padding')
    stride = utils.convert_to_list(stride, 2, 'stride')
    dilation = utils.convert_to_list(dilation, 2, 'dilation')
G
guosheng 已提交
2692

C
chengduoZH 已提交
2693 2694
    if not isinstance(use_cudnn, bool):
        raise ValueError("use_cudnn should be True or False")
G
guosheng 已提交
2695

Y
Yu Yang 已提交
2696 2697 2698 2699 2700
    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]
G
guosheng 已提交
2701

Y
Yu Yang 已提交
2702 2703
        h_in = input.shape[2]
        w_in = input.shape[3]
G
guosheng 已提交
2704

C
chengduoZH 已提交
2705
        filter_size_h = (output_size[0] - (h_in - 1) * stride[0] + 2 *
M
minqiyang 已提交
2706
                         padding[0] - 1) // dilation[0] + 1
C
chengduoZH 已提交
2707
        filter_size_w = (output_size[1] - (w_in - 1) * stride[1] + 2 *
M
minqiyang 已提交
2708
                         padding[1] - 1) // dilation[1] + 1
Y
Yu Yang 已提交
2709
        filter_size = [filter_size_h, filter_size_w]
C
chengduoZH 已提交
2710 2711 2712
    else:
        filter_size = utils.convert_to_list(filter_size, 2,
                                            'conv2d_transpose.filter_size')
C
chengduo 已提交
2713

2714 2715 2716 2717 2718 2719 2720
    if output_size is None:
        output_size = []
    elif isinstance(output_size, list) or isinstance(output_size, int):
        output_size = utils.convert_to_list(output_size, 2, 'output_size')
    else:
        raise ValueError("output_size should be list or int")
    padding = utils.convert_to_list(padding, 2, 'padding')
2721
    groups = 1 if groups is None else groups
M
minqiyang 已提交
2722
    filter_shape = [input_channel, num_filters // groups] + filter_size
C
chengduo 已提交
2723

Y
Yu Yang 已提交
2724 2725 2726
    img_filter = helper.create_parameter(
        dtype=input.dtype, shape=filter_shape, attr=helper.param_attr)

X
Xin Pan 已提交
2727
    pre_bias = helper.create_variable_for_type_inference(dtype=input.dtype)
Y
Yu Yang 已提交
2728
    helper.append_op(
2729
        type=op_type,
Y
Yu Yang 已提交
2730 2731
        inputs={'Input': [input],
                'Filter': [img_filter]},
2732
        outputs={'Output': pre_bias},
C
chengduoZH 已提交
2733
        attrs={
2734
            'output_size': output_size,
2735 2736 2737 2738 2739
            'strides': stride,
            'paddings': padding,
            'dilations': dilation,
            'groups': groups,
            'use_cudnn': use_cudnn
Y
Yu Yang 已提交
2740 2741
        })

2742 2743 2744
    pre_act = helper.append_bias_op(pre_bias, dim_start=1, dim_end=2)
    out = helper.append_activation(pre_act)
    return out
Y
Yu Yang 已提交
2745 2746


2747
def conv3d_transpose(input,
Y
Yu Yang 已提交
2748 2749 2750
                     num_filters,
                     output_size=None,
                     filter_size=None,
C
chengduoZH 已提交
2751 2752 2753
                     padding=0,
                     stride=1,
                     dilation=1,
2754
                     groups=None,
C
caoying03 已提交
2755
                     param_attr=None,
2756
                     bias_attr=None,
C
chengduoZH 已提交
2757
                     use_cudnn=True,
2758
                     act=None,
C
caoying03 已提交
2759
                     name=None):
Y
Yu Yang 已提交
2760
    """
2761
    **Convlution3D transpose layer**
2762

2763
    The convolution3D transpose layer calculates the output based on the input,
2764
    filter, and dilations, strides, paddings. Input(Input) and output(Output)
2765 2766 2767 2768 2769 2770
    are in NCDHW format. Where N is batch size, C is the number of channels,
    D is the depth of the feature, H is the height of the feature, and W
    is the width of the feature. Parameters(dilations, strides, paddings) are
    two elements. These two elements represent height and width, respectively.
    The details of convolution transpose layer, please refer to the following
    explanation and references `therein <http://www.matthewzeiler.com/wp-content/uploads/2017/07/cvpr2010.pdf>`_.
2771 2772 2773
    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.
2774 2775 2776 2777 2778

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

    .. math::

2779
        Out = \sigma (W \\ast X + b)
2780 2781 2782

    In the above equation:

2783 2784
    * :math:`X`: Input value, a tensor with NCDHW format.
    * :math:`W`: Filter value, a tensor with MCDHW format.
2785 2786 2787 2788
    * :math:`\\ast`: Convolution operation.
    * :math:`b`: Bias value, a 2-D tensor with shape [M, 1].
    * :math:`\\sigma`: Activation function.
    * :math:`Out`: Output value, the shape of :math:`Out` and :math:`X` may be different.
Y
Yu Yang 已提交
2789

2790 2791 2792 2793
    Example:

        - Input:

2794
          Input shape: :math:`(N, C_{in}, D_{in}, H_{in}, W_{in})`
2795

2796
          Filter shape: :math:`(C_{in}, C_{out}, D_f, H_f, W_f)`
2797 2798 2799

        - Output:

2800
          Output shape: :math:`(N, C_{out}, D_{out}, H_{out}, W_{out})`
2801 2802

        Where
Y
Yu Yang 已提交
2803

2804 2805
        .. math::

2806 2807 2808
           D_{out} &= (D_{in} - 1) * strides[0] - 2 * paddings[0] + dilations[0] * (D_f - 1) + 1 \\\\
           H_{out} &= (H_{in} - 1) * strides[1] - 2 * paddings[1] + dilations[1] * (H_f - 1) + 1 \\\\
           W_{out} &= (W_{in} - 1) * strides[2] - 2 * paddings[2] + dilations[2] * (W_f - 1) + 1
Y
Yu Yang 已提交
2809 2810

    Args:
2811
        input(Variable): The input image with [N, C, D, H, W] format.
2812 2813 2814
        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
2815
            tuple, it must contain three integers, (image_D, image_H, image_W). This
2816 2817
            parameter only works when filter_size is None.
        filter_size(int|tuple|None): The filter size. If filter_size is a tuple,
2818
            it must contain three integers, (filter_size_D, filter_size_H, filter_size_W).
2819 2820 2821
            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
2822 2823
            contain three integers, (padding_D, padding_H, padding_W). Otherwise, the
            padding_D = padding_H = padding_W = padding. Default: padding = 0.
2824
        stride(int|tuple): The stride size. If stride is a tuple, it must
2825 2826
            contain three integers, (stride_D, stride_H, stride_W). Otherwise, the
            stride_D = stride_H = stride_W = stride. Default: stride = 1.
2827
        dilation(int|tuple): The dilation size. If dilation is a tuple, it must
2828 2829 2830
            contain three integers, (dilation_D, dilation_H, dilation_W). Otherwise, the
            dilation_D = dilation_H = dilation_W = dilation. Default: dilation = 1.
        groups(int): The groups number of the Conv3d transpose layer. Inspired by
2831 2832 2833 2834 2835
            grouped convolution in Alex Krizhevsky's Deep CNN paper, in which
            when group=2, the first half of the filters is only connected to the
            first half of the input channels, while the second half of the
            filters is only connected to the second half of the input channels.
            Default: groups=1
C
chengduo 已提交
2836 2837 2838 2839 2840 2841 2842 2843 2844
        param_attr (ParamAttr|None): The parameter attribute for learnable parameters/weights
            of conv3d_transpose. If it is set to None or one attribute of ParamAttr, conv3d_transpose
            will create ParamAttr as param_attr. If the Initializer of the param_attr
            is not set, the parameter is initialized with Xavier. Default: None.
        bias_attr (ParamAttr|bool|None): The parameter attribute for the bias of conv3d_transpose.
            If it is set to False, no bias will be added to the output units.
            If it is set to None or one attribute of ParamAttr, conv3d_transpose
            will create ParamAttr as bias_attr. If the Initializer of the bias_attr
            is not set, the bias is initialized zero. Default: None.
2845 2846
        use_cudnn(bool): Use cudnn kernel or not, it is valid only when the cudnn
            library is installed. Default: True
C
chengduo 已提交
2847 2848
        act (str): Activation type, if it is set to None, activation is not appended.
            Default: None.
2849 2850
        name(str|None): A name for this layer(optional). If set None, the layer
            will be named automatically.
Y
Yu Yang 已提交
2851 2852

    Returns:
2853
        Variable: The tensor variable storing the convolution transpose result.
2854 2855

    Raises:
2856 2857
        ValueError: If the shapes of input, filter_size, stride, padding and
                    groups mismatch.
2858 2859 2860 2861

    Examples:
       .. code-block:: python

2862 2863
          data = fluid.layers.data(name='data', shape=[3, 12, 32, 32], dtype='float32')
          conv3d_transpose = fluid.layers.conv3d_transpose(input=data, num_filters=2, filter_size=3)
Y
Yu Yang 已提交
2864
    """
C
chengduo 已提交
2865
    assert param_attr is not False, "param_attr should not be False in conv3d_transpose."
2866 2867
    l_type = "conv3d_transpose"
    helper = LayerHelper(l_type, **locals())
Y
Yu Yang 已提交
2868
    if not isinstance(input, Variable):
2869
        raise TypeError("Input of conv3d_transpose must be Variable")
Y
Yu Yang 已提交
2870 2871
    input_channel = input.shape[1]

2872 2873 2874
    padding = utils.convert_to_list(padding, 3, 'padding')
    stride = utils.convert_to_list(stride, 3, 'stride')
    dilation = utils.convert_to_list(dilation, 3, 'dilation')
C
chengduoZH 已提交
2875

C
chengduoZH 已提交
2876 2877 2878
    if not isinstance(use_cudnn, bool):
        raise ValueError("use_cudnn should be True or False")

Y
Yu Yang 已提交
2879 2880 2881 2882 2883 2884
    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]

2885 2886 2887
        d_in = input.shape[2]
        h_in = input.shape[3]
        w_in = input.shape[4]
C
chengduoZH 已提交
2888

2889
        filter_size_d = (output_size[0] - (d_in - 1) * stride[0] + 2 *
M
minqiyang 已提交
2890
                         padding[0] - 1) // dilation[0] + 1
2891
        filter_size_h = (output_size[1] - (h_in - 1) * stride[1] + 2 *
M
minqiyang 已提交
2892
                         padding[1] - 1) // dilation[1] + 1
2893
        filter_size_w = (output_size[2] - (w_in - 1) * stride[2] + 2 *
M
minqiyang 已提交
2894
                         padding[2] - 1) // dilation[2] + 1
2895
        filter_size = [filter_size_d, filter_size_h, filter_size_w]
C
chengduoZH 已提交
2896
    else:
2897 2898
        filter_size = utils.convert_to_list(filter_size, 3,
                                            'conv3d_transpose.filter_size')
Y
Yu Yang 已提交
2899

2900
    groups = 1 if groups is None else groups
M
minqiyang 已提交
2901
    filter_shape = [input_channel, num_filters // groups] + filter_size
Y
Yu Yang 已提交
2902 2903 2904
    img_filter = helper.create_parameter(
        dtype=input.dtype, shape=filter_shape, attr=helper.param_attr)

X
Xin Pan 已提交
2905
    pre_bias = helper.create_variable_for_type_inference(dtype=input.dtype)
Y
Yu Yang 已提交
2906
    helper.append_op(
2907
        type=l_type,
Y
Yu Yang 已提交
2908 2909
        inputs={'Input': [input],
                'Filter': [img_filter]},
2910
        outputs={'Output': pre_bias},
C
chengduoZH 已提交
2911 2912 2913 2914
        attrs={
            'strides': stride,
            'paddings': padding,
            'dilations': dilation,
2915
            'groups': groups,
C
chengduoZH 已提交
2916 2917
            'use_cudnn': use_cudnn
        })
Y
Yu Yang 已提交
2918

2919 2920
    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
2921
    return out
Y
yangyaming 已提交
2922 2923


Y
yangyaming 已提交
2924
def sequence_expand(x, y, ref_level=-1, name=None):
2925
    """Sequence Expand Layer. This layer will expand the input variable **x**
Y
yangyaming 已提交
2926 2927 2928 2929
    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:
2930 2931 2932 2933 2934

    .. code-block:: text

        * Case 1
            x is a LoDTensor:
2935
                x.lod  = [[2,        2]]
Y
yangyaming 已提交
2936
                x.data = [[a], [b], [c], [d]]
2937 2938 2939
                x.dims = [4, 1]

            y is a LoDTensor:
2940 2941
                y.lod = [[2,    2],
                         [3, 3, 1, 1]]
2942

Y
yangyaming 已提交
2943
            ref_level: 0
2944

Y
yangyaming 已提交
2945
            then output is a 1-level LoDTensor:
2946
                out.lod =  [[2,        2,        2,        2]]
Y
yangyaming 已提交
2947
                out.data = [[a], [b], [a], [b], [c], [d], [c], [d]]
2948 2949 2950 2951
                out.dims = [8, 1]

        * Case 2
            x is a Tensor:
Y
yangyaming 已提交
2952
                x.data = [[a], [b], [c]]
2953 2954 2955
                x.dims = [3, 1]

            y is a LoDTensor:
2956
                y.lod = [[2, 0, 3]]
2957

Y
yangyaming 已提交
2958
            ref_level: -1
2959

Y
yangyaming 已提交
2960 2961 2962
            then output is a Tensor:
                out.data = [[a], [a], [c], [c], [c]]
                out.dims = [5, 1]
2963 2964 2965
    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
2966 2967
        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
2968
        name(str|None): A name for this layer(optional). If set None, the layer
Y
yangyaming 已提交
2969
                        will be named automatically.
2970 2971 2972 2973 2974 2975 2976 2977 2978 2979

    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
2980
            out = layers.sequence_expand(x=x, y=y, ref_level=0)
2981
    """
Y
yangyaming 已提交
2982
    helper = LayerHelper('sequence_expand', input=x, **locals())
2983
    dtype = helper.input_dtype()
X
Xin Pan 已提交
2984
    tmp = helper.create_variable_for_type_inference(dtype)
2985
    helper.append_op(
Y
yangyaming 已提交
2986 2987 2988 2989 2990
        type='sequence_expand',
        inputs={'X': x,
                'Y': y},
        outputs={'Out': tmp},
        attrs={'ref_level': ref_level})
2991
    return tmp
2992 2993


C
chengduo 已提交
2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049
def sequence_expand_as(x, y, name=None):
    """Sequence Expand As Layer. This layer will expand the input variable **x**
    according to the zeroth level lod of **y**. Current implementation requires
    the level number of Input(Y)'s lod must be 1, and the first dimension of
    Input(X) should be equal to the size of Input(Y)'s zeroth level lod, and
    lod of Input(X) is not considered.

    Following examples will explain how sequence_expand_as works:

    .. code-block:: text

        * Case 1:

            Given a 1-level LoDTensor input(X)
                X.data = [[a], [b], [c], [d]]
                X.dims = [4, 1]
            and input(Y)
                Y.lod = [[0, 3, 6, 7, 8]]
            ref_level: 0
            then we get 1-level LoDTensor
                Out.lod =  [[0,            3,              6,  7,  8]]
                Out.data = [[a], [a], [a], [b], [b], [b], [c], [d]]
                Out.dims = [8, 1]

        * Case 2:

            Given a common Tensor input(X)
                X.data = [[a, b], [c, d], [e, f]]
                X.dims = [3, 2]
            and input(Y)
                Y.lod = [[0, 2, 3, 6]]
            ref_level: 0
            then we get a common LoDTensor
                Out.lod =  [[0,             2,     3,                    6]]
                Out.data = [[a, b], [a, b] [c, d], [e, f], [e, f], [e, f]]
                Out.dims = [6, 2]

    Args:
        x (Variable): The input variable which is a Tensor or LoDTensor.
        y (Variable): The input variable which is a LoDTensor.
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.

    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)
            out = layers.sequence_expand_as(x=x, y=y)
    """
    helper = LayerHelper('sequence_expand_as', input=x, **locals())
    dtype = helper.input_dtype()
X
Xin Pan 已提交
3050
    tmp = helper.create_variable_for_type_inference(dtype)
C
chengduo 已提交
3051 3052 3053 3054 3055 3056 3057 3058
    helper.append_op(
        type='sequence_expand_as',
        inputs={'X': x,
                'Y': y},
        outputs={'Out': tmp})
    return tmp


F
fengjiayi 已提交
3059
@templatedoc()
3060
def sequence_pad(x, pad_value, maxlen=None, name=None):
F
fengjiayi 已提交
3061 3062 3063 3064 3065
    """
    ${comment}

    Args:
        x(Variable): Input variable which should contain lod information.
M
minqiyang 已提交
3066 3067 3068
        pad_value(Variable): The Variable that holds values that will be fill
            into padded steps. It can be a scalar or a tensor whose shape
            equals to time steps in sequences. If it's a scalar, it will be
F
fengjiayi 已提交
3069
            automatically broadcasted to the shape of time step.
M
minqiyang 已提交
3070 3071 3072 3073
        maxlen(int, default None): The length of padded sequences. It can be
            None or any positive int. When it is None, all sequences will be
            padded up to the length of the longest one among them; when it a
            certain positive value, it must be greater than the length of the
3074 3075 3076
            longest original sequence.
        name(str|None): A name for this layer(optional). If set None, the layer
            will be named automatically.
M
minqiyang 已提交
3077

F
fengjiayi 已提交
3078
    Returns:
M
minqiyang 已提交
3079
        Variable: The padded sequence batch and the original lengths before
3080
                  padding. All sequences has the same length.
M
minqiyang 已提交
3081

F
fengjiayi 已提交
3082 3083 3084 3085 3086 3087 3088
    Examples:
        .. code-block:: python

            import numpy

            x = fluid.layers.data(name='y', shape=[10, 5],
                             dtype='float32', lod_level=1)
G
gmcather 已提交
3089
            pad_value = fluid.layers.assign(
D
dongzhihong 已提交
3090
                input=numpy.array([0.0], dtype=numpy.float32))
F
fengjiayi 已提交
3091 3092 3093 3094 3095
            out = fluid.layers.sequence_pad(x=x, pad_value=pad_value)
    """

    helper = LayerHelper('sequence_pad', input=x, **locals())
    dtype = helper.input_dtype()
X
Xin Pan 已提交
3096 3097
    out = helper.create_variable_for_type_inference(dtype)
    length = helper.create_variable_for_type_inference(dtype)
3098 3099 3100 3101

    pad_value.stop_gradient = True
    length.stop_gradient = True

F
fengjiayi 已提交
3102 3103 3104 3105 3106 3107
    if maxlen is None:
        maxlen = -1
    helper.append_op(
        type='sequence_pad',
        inputs={'X': x,
                'PadValue': pad_value},
3108 3109
        outputs={'Out': out,
                 'Length': length},
F
fengjiayi 已提交
3110
        attrs={'padded_length': maxlen})
3111
    return out, length
F
fengjiayi 已提交
3112 3113


3114
def sequence_unpad(x, length, name=None):
Y
Yibing Liu 已提交
3115
    """
3116
    **Sequence Unpad Layer**
Y
Yibing Liu 已提交
3117

3118 3119
    This layer removes the padding data in the input sequences and convert
    them into sequences with actual length as output, identitied by lod
Y
Yibing Liu 已提交
3120 3121 3122 3123 3124 3125 3126 3127 3128
    information.

    .. code-block:: text

	Example:

	Given input Variable **x**:
	    x.data = [[ 1.0,  2.0,  3.0,  4.0,  5.0],
		      [ 6.0,  7.0,  8.0,  9.0, 10.0],
3129 3130 3131
		      [11.0, 12.0, 13.0, 14.0, 15.0]],

	in which there are 3 sequences padded to length 5, and the acutal length
3132
	specified by input Variable **length**:
Y
Yibing Liu 已提交
3133 3134 3135 3136 3137 3138

	    length.data = [[2], [3], [4]],

	after unpadding, the output Variable will be:

	    out.data = [[1.0, 2.0, 6.0, 7.0, 8.0, 11.0, 12.0, 13.0, 14.0]]
3139
	    out.lod = [[2, 3, 4]]
Y
Yibing Liu 已提交
3140 3141 3142 3143 3144 3145

    Args:
        x(Variable): Input Variable which contains the padded sequences with
            equal length.
        length(Variable): The Variable that specifies the actual ength of
            sequences after unpadding.
3146 3147
        name(str|None): A name for this layer(optional). If set None, the layer
            will be named automatically.
Y
Yibing Liu 已提交
3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161

    Returns:
        Variable: The Variable contains the unpadded sequences.

    Examples:
        .. code-block:: python

            x = fluid.layers.data(name='x', shape=[10, 5], dtype='float32')
            len = fluid.layers.data(name='length', shape=[1], dtype='int64')
            out = fluid.layers.sequence_unpad(x=x, length=len)
    """

    helper = LayerHelper('sequence_unpad', input=x, **locals())
    dtype = helper.input_dtype()
X
Xin Pan 已提交
3162
    out = helper.create_variable_for_type_inference(dtype)
Y
Yibing Liu 已提交
3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173

    length.stop_gradient = True

    helper.append_op(
        type='sequence_unpad',
        inputs={'X': x,
                'Length': length},
        outputs={'Out': out})
    return out


3174 3175 3176 3177 3178 3179 3180 3181 3182
def beam_search(pre_ids,
                pre_scores,
                ids,
                scores,
                beam_size,
                end_id,
                level=0,
                name=None):
    """
3183 3184
    Beam search is a classical algorithm for selecting candidate words in a
    machine translation task.
Y
Yan Chunwei 已提交
3185 3186 3187

    Refer to `Beam search <https://en.wikipedia.org/wiki/Beam_search>`_
    for more details.
M
minqiyang 已提交
3188 3189

    This layer does the search in beams for one time step. Specifically, it
3190 3191 3192 3193 3194 3195
    selects the top-K candidate word ids of current step from :attr:`ids`
    according to their :attr:`scores` for all source sentences, where K is
    :attr:`beam_size` and :attr:`ids, scores` are predicted results from the
    computation cell. Additionally, :attr:`pre_ids` and :attr:`pre_scores` are
    the output of beam_search at previous step, they are needed for special use
    to handle ended candidate translations.
M
minqiyang 已提交
3196

3197 3198 3199 3200 3201 3202 3203 3204
    Note that the :attr:`scores` passed in should be accumulated scores, and
    length penalty should be done with extra operators before calculating the
    accumulated scores if needed, also suggest finding top-K before it and
    using the top-K candidates following.

    Please see the following demo for a fully beam search usage example:

        fluid/tests/book/test_machine_translation.py
Y
Yan Chunwei 已提交
3205

3206
    Args:
3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231
        pre_ids(Variable): The LodTensor variable which is the output of
            beam_search at previous step. It should be a LodTensor with shape
            :math:`(batch_size, 1)` and lod
            :math:`[[0, 1, ... , batch_size], [0, 1, ..., batch_size]]` at the
            first step.
        pre_scores(Variable): The LodTensor variable which is the output of
            beam_search at previous step.
        ids(Variable): The LodTensor variable containing the candidates ids.
            Its shape should be :math:`(batch_size \\times beam_size, K)`,
            where :math:`K` supposed to be :attr:`beam_size`.
        scores(Variable): The LodTensor variable containing the accumulated
            scores corresponding to :attr:`ids` and its shape is the same as
            the shape of :attr:`ids`.
        beam_size(int): The beam width used in beam search.
        end_id(int): The id of end token.
        level(int, default 0): It can be ignored and mustn't change currently.
            It means the source level of lod, which is explained as following.
            The lod level of :attr:`ids` should be 2. The first level is source
            level which describes how many prefixes (branchs) for each source
            sentece (beam), and the second level is sentence level which
            describes how these candidates belong to the prefix. The paths
            linking prefixes and selected candidates are organized and reserved
            in lod.
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.
F
fengjiayi 已提交
3232

3233
    Returns:
3234 3235
        Variable: The LodTensor pair containing the selected ids and the \
            corresponding scores.
Y
Yan Chunwei 已提交
3236 3237 3238 3239

    Examples:
        .. code-block:: python

3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256
            # Suppose `probs` contains predicted results from the computation
            # cell and `pre_ids` and `pre_scores` is the output of beam_search
            # at previous step.
            topk_scores, topk_indices = layers.topk(probs, k=beam_size)
            accu_scores = layers.elementwise_add(
                x=layers.log(x=topk_scores)),
                y=layers.reshape(
                    pre_scores, shape=[-1]),
                axis=0)
            selected_ids, selected_scores = layers.beam_search(
                pre_ids=pre_ids,
                pre_scores=pre_scores,
                ids=topk_indices,
                scores=accu_scores,
                beam_size=beam_size,
                end_id=end_id)
    """
Q
Qiao Longfei 已提交
3257 3258 3259 3260
    helper = LayerHelper('beam_search', **locals())
    score_type = scores.dtype
    id_type = ids.dtype

X
Xin Pan 已提交
3261 3262 3263
    selected_scores = helper.create_variable_for_type_inference(
        dtype=score_type)
    selected_ids = helper.create_variable_for_type_inference(dtype=id_type)
Q
Qiao Longfei 已提交
3264 3265 3266 3267 3268

    helper.append_op(
        type='beam_search',
        inputs={
            'pre_ids': pre_ids,
3269
            'pre_scores': pre_scores,
Q
Qiao Longfei 已提交
3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286
            '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


3287 3288 3289 3290 3291 3292 3293
def beam_search_decode(ids, scores, beam_size, end_id, name=None):
    """
    Beam Search Decode Layer. This layer constructs the full hypotheses for
    each source sentence by walking back along the LoDTensorArray :attr:`ids`
    whose lods can be used to restore the path in the beam search tree.
    Please see the following demo for a fully beam search usage example:
        fluid/tests/book/test_machine_translation.py
G
guosheng 已提交
3294

3295 3296 3297 3298 3299 3300 3301 3302 3303
    Args:
        ids(Variable): The LodTensorArray variable containing the selected ids
            of all steps.
        scores(Variable): The LodTensorArray variable containing the selected
            scores of all steps.
        beam_size(int): The beam width used in beam search.
        end_id(int): The id of end token.
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.
G
guosheng 已提交
3304

3305 3306 3307 3308 3309 3310
    Returns:
        Variable: The LodTensor pair containing the generated id sequences \
            and the corresponding scores. The shapes and lods of the two \
            LodTensor are same. The lod level is 2 and the two levels \
            separately indicate how many hypotheses each source sentence has \
            and how many ids each hypothesis has.
G
guosheng 已提交
3311

3312 3313 3314 3315 3316 3317 3318 3319
    Examples:
        .. code-block:: python
            # Suppose `ids` and `scores` are LodTensorArray variables reserving
            # the selected ids and scores of all steps
            finished_ids, finished_scores = layers.beam_search_decode(
                ids, scores, beam_size=5, end_id=0)
    """
    helper = LayerHelper('beam_search_decode', **locals())
X
Xin Pan 已提交
3320 3321
    sentence_ids = helper.create_variable_for_type_inference(dtype=ids.dtype)
    sentence_scores = helper.create_variable_for_type_inference(dtype=ids.dtype)
3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336

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

    return sentence_ids, sentence_scores


Y
yangyaming 已提交
3337 3338 3339 3340
def lstm_unit(x_t,
              hidden_t_prev,
              cell_t_prev,
              forget_bias=0.0,
Y
yangyaming 已提交
3341
              param_attr=None,
C
caoying03 已提交
3342 3343
              bias_attr=None,
              name=None):
Y
yangyaming 已提交
3344 3345 3346 3347
    """Lstm unit layer. The equation of a lstm step is:

        .. math::

3348
            i_t & = \sigma(W_{x_i}x_{t} + W_{h_i}h_{t-1} + b_i)
Y
yangyaming 已提交
3349

3350
            f_t & = \sigma(W_{x_f}x_{t} + W_{h_f}h_{t-1} + b_f)
Y
yangyaming 已提交
3351

3352
            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
3353

3354
            o_t & = \sigma(W_{x_o}x_{t} + W_{h_o}h_{t-1} + b_o)
Y
yangyaming 已提交
3355 3356 3357

            h_t & = o_t tanh(c_t)

3358 3359 3360 3361 3362 3363
    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
3364 3365 3366

        .. math::

3367
            L_{i_t} = W_{x_i}x_{t} + W_{h_i}h_{t-1} + b_i
Y
yangyaming 已提交
3368 3369 3370 3371 3372 3373 3374 3375

    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
3376
    This layer has two outputs including :math:`h_t` and :math:`o_t`.
Y
yangyaming 已提交
3377 3378

    Args:
Y
yangyaming 已提交
3379 3380 3381 3382 3383 3384
        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
3385
        forget_bias (float): The forget bias of lstm unit.
C
chengduo 已提交
3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397
        param_attr(ParamAttr|None): The parameter attribute for the learnable
                               hidden-hidden weights.
                               If it is set to None or one attribute of ParamAttr,
                               lstm_unit will create ParamAttr as param_attr.
                               If the Initializer of the param_attr is not set, the
                               parameter is initialized with Xavier. Default: None.
        bias_attr (ParamAttr|None): The bias attribute for the learnable bias
                              weights. If it is set to False, no bias will be added
                              to the output units. If it is set to None or one attribute of ParamAttr,
                              lstm_unit will create ParamAttr as bias_attr.
                              If the Initializer of the bias_attr is not set,
                              the bias is initialized zero. Default: None.
C
caoying03 已提交
3398 3399
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
Y
yangyaming 已提交
3400 3401

    Returns:
Y
yangyaming 已提交
3402
        tuple: The hidden value and cell value of lstm unit.
Y
yangyaming 已提交
3403 3404

    Raises:
3405 3406 3407 3408
        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
3409 3410 3411 3412 3413 3414

    Examples:

        .. code-block:: python

             x_t = fluid.layers.fc(input=x_t_data, size=10)
3415
             prev_hidden = fluid.layers.fc(input=prev_hidden_data, size=30)
Y
yangyaming 已提交
3416
             prev_cell = fluid.layers.fc(input=prev_cell_data, size=30)
Y
yangyaming 已提交
3417
             hidden_value, cell_value = fluid.layers.lstm_unit(x_t=x_t,
Y
yangyaming 已提交
3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433
                                                    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
3434
        raise ValueError("The 1st dimensions of x_t, hidden_t_prev and "
3435 3436 3437 3438
                         "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
3439 3440
                         "cell_t_prev must be the same.")

Y
yangyaming 已提交
3441 3442 3443
    if bias_attr is None:
        bias_attr = ParamAttr()

Y
yangyaming 已提交
3444
    size = cell_t_prev.shape[1]
3445
    concat_out = concat(input=[x_t, hidden_t_prev], axis=1)
Y
yangyaming 已提交
3446 3447
    fc_out = fc(input=concat_out,
                size=4 * size,
Y
yangyaming 已提交
3448
                param_attr=param_attr,
3449
                bias_attr=bias_attr)
Y
yangyaming 已提交
3450
    dtype = x_t.dtype
X
Xin Pan 已提交
3451 3452
    c = helper.create_variable_for_type_inference(dtype)
    h = helper.create_variable_for_type_inference(dtype)
Y
yangyaming 已提交
3453 3454 3455 3456 3457 3458 3459 3460 3461

    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
3462
    return h, c
G
guosheng 已提交
3463 3464


C
caoying03 已提交
3465
def reduce_sum(input, dim=None, keep_dim=False, name=None):
G
guosheng 已提交
3466
    """
Y
yangyaming 已提交
3467
    Computes the sum of tensor elements over the given dimension.
G
guosheng 已提交
3468 3469 3470

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
W
whs 已提交
3471
        dim (list|int|None): The dimensions along which the sum is performed. If
Y
yangyaming 已提交
3472 3473
            :attr:`None`, sum all elements of :attr:`input` and return a
            Tensor variable with a single element, otherwise must be in the
W
whs 已提交
3474 3475
            range :math:`[-rank(input), rank(input))`. If :math:`dim[i] < 0`,
            the dimension to reduce is :math:`rank + dim[i]`.
3476
        keep_dim (bool|False): Whether to reserve the reduced dimension in the
Y
yangyaming 已提交
3477
            output Tensor. The result tensor will have one fewer dimension
G
guosheng 已提交
3478
            than the :attr:`input` unless :attr:`keep_dim` is true.
C
caoying03 已提交
3479 3480
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
G
guosheng 已提交
3481 3482 3483

    Returns:
        Variable: The reduced Tensor variable.
F
fengjiayi 已提交
3484

G
guosheng 已提交
3485 3486 3487 3488 3489 3490
    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]]
Q
qiaolongfei 已提交
3491
            # Each example is followed by the corresponding output tensor.
G
guosheng 已提交
3492 3493 3494 3495
            fluid.layers.reduce_sum(x)  # [3.5]
            fluid.layers.reduce_sum(x, dim=0)  # [0.3, 0.5, 1.1, 1.6]
            fluid.layers.reduce_sum(x, dim=-1)  # [1.9, 1.6]
            fluid.layers.reduce_sum(x, dim=1, keep_dim=True)  # [[1.9], [1.6]]
W
whs 已提交
3496 3497 3498 3499

            # x is a Tensor variable with shape [2, 2, 2] and elements as below:
            #      [[[1, 2], [3, 4]],
            #      [[5, 6], [7, 8]]]
Q
qiaolongfei 已提交
3500
            # Each example is followed by the corresponding output tensor.
W
whs 已提交
3501 3502 3503
            fluid.layers.reduce_sum(x, dim=[1, 2]) # [10, 26]
            fluid.layers.reduce_sum(x, dim=[0, 1]) # [16, 20]

G
guosheng 已提交
3504 3505
    """
    helper = LayerHelper('reduce_sum', **locals())
X
Xin Pan 已提交
3506
    out = helper.create_variable_for_type_inference(dtype=helper.input_dtype())
W
whs 已提交
3507 3508
    if dim is not None and not isinstance(dim, list):
        dim = [dim]
G
guosheng 已提交
3509 3510 3511 3512 3513
    helper.append_op(
        type='reduce_sum',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
W
whs 已提交
3514
            'dim': dim if dim != None else [0],
G
guosheng 已提交
3515 3516 3517 3518
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out
G
guosheng 已提交
3519 3520


C
caoying03 已提交
3521
def reduce_mean(input, dim=None, keep_dim=False, name=None):
G
guosheng 已提交
3522
    """
Y
Yibing Liu 已提交
3523
    Computes the mean of the input tensor's elements along the given dimension.
G
guosheng 已提交
3524 3525 3526

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
Y
Yibing Liu 已提交
3527 3528 3529
        dim (list|int|None): The dimension along which the mean is computed. If
            `None`, compute the mean over all elements of :attr:`input`
            and return a variable with a single element, otherwise it
Y
yangyaming 已提交
3530
            must be in the range :math:`[-rank(input), rank(input))`. If
3531
            :math:`dim[i] < 0`, the dimension to reduce is
Y
Yibing Liu 已提交
3532
            :math:`rank(input) + dim[i]`.
Y
yangyaming 已提交
3533 3534
        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
3535
            than the :attr:`input` unless :attr:`keep_dim` is true.
Y
Yibing Liu 已提交
3536
        name(str|None): A name for this layer(optional). If set `None`, the layer
C
caoying03 已提交
3537
                       will be named automatically.
G
guosheng 已提交
3538 3539

    Returns:
Y
Yibing Liu 已提交
3540
        Variable: The reduced mean Variable.
F
fengjiayi 已提交
3541

G
guosheng 已提交
3542 3543 3544 3545 3546 3547 3548 3549 3550 3551
    Examples:
        .. code-block:: python

            # x is a Tensor variable with following elements:
            #    [[0.2, 0.3, 0.5, 0.9]
            #     [0.1, 0.2, 0.6, 0.7]]
            # Each example is followed by the correspending output tensor.
            fluid.layers.reduce_mean(x)  # [0.4375]
            fluid.layers.reduce_mean(x, dim=0)  # [0.15, 0.25, 0.55, 0.8]
            fluid.layers.reduce_mean(x, dim=-1)  # [0.475, 0.4]
F
stash  
fengjiayi 已提交
3552 3553
            fluid.layers.reduce_mean(
                x, dim=1, keep_dim=True)  # [[0.475], [0.4]]
W
whs 已提交
3554 3555 3556 3557 3558 3559 3560

            # x is a Tensor variable with shape [2, 2, 2] and elements as below:
            #      [[[1.0, 2.0], [3.0, 4.0]],
            #      [[5.0, 6.0], [7.0, 8.0]]]
            # Each example is followed by the correspending output tensor.
            fluid.layers.reduce_mean(x, dim=[1, 2]) # [2.5, 6.5]
            fluid.layers.reduce_mean(x, dim=[0, 1]) # [4.0, 5.0]
G
guosheng 已提交
3561 3562
    """
    helper = LayerHelper('reduce_mean', **locals())
X
Xin Pan 已提交
3563
    out = helper.create_variable_for_type_inference(dtype=helper.input_dtype())
W
whs 已提交
3564 3565
    if dim is not None and not isinstance(dim, list):
        dim = [dim]
G
guosheng 已提交
3566 3567 3568 3569 3570
    helper.append_op(
        type='reduce_mean',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
W
whs 已提交
3571
            'dim': dim if dim != None else [0],
G
guosheng 已提交
3572 3573 3574 3575
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out
3576 3577


C
caoying03 已提交
3578
def reduce_max(input, dim=None, keep_dim=False, name=None):
3579
    """
Y
yangyaming 已提交
3580
    Computes the maximum of tensor elements over the given dimension.
3581 3582 3583

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
W
whs 已提交
3584
        dim (list|int|None): The dimension along which the maximum is computed.
Y
yangyaming 已提交
3585 3586 3587
            If :attr:`None`, compute the maximum over all elements of
            :attr:`input` and return a Tensor variable with a single element,
            otherwise must be in the range :math:`[-rank(input), rank(input))`.
W
whs 已提交
3588
            If :math:`dim[i] < 0`, the dimension to reduce is :math:`rank + dim[i]`.
Y
yangyaming 已提交
3589 3590
        keep_dim (bool): Whether to reserve the reduced dimension in the
            output Tensor. The result tensor will have one fewer dimension
3591
            than the :attr:`input` unless :attr:`keep_dim` is true.
C
caoying03 已提交
3592 3593
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
3594 3595 3596

    Returns:
        Variable: The reduced Tensor variable.
Y
yangyaming 已提交
3597

3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608
    Examples:
        .. code-block:: python

            # x is a Tensor variable with following elements:
            #    [[0.2, 0.3, 0.5, 0.9]
            #     [0.1, 0.2, 0.6, 0.7]]
            # Each example is followed by the correspending output tensor.
            fluid.layers.reduce_max(x)  # [0.9]
            fluid.layers.reduce_max(x, dim=0)  # [0.2, 0.3, 0.6, 0.9]
            fluid.layers.reduce_max(x, dim=-1)  # [0.9, 0.7]
            fluid.layers.reduce_max(x, dim=1, keep_dim=True)  # [[0.9], [0.7]]
W
whs 已提交
3609 3610 3611 3612 3613 3614 3615

            # x is a Tensor variable with shape [2, 2, 2] and elements as below:
            #      [[[1.0, 2.0], [3.0, 4.0]],
            #      [[5.0, 6.0], [7.0, 8.0]]]
            # Each example is followed by the correspending output tensor.
            fluid.layers.reduce_max(x, dim=[1, 2]) # [4.0, 8.0]
            fluid.layers.reduce_max(x, dim=[0, 1]) # [7.0, 8.0]
3616 3617
    """
    helper = LayerHelper('reduce_max', **locals())
X
Xin Pan 已提交
3618
    out = helper.create_variable_for_type_inference(dtype=helper.input_dtype())
W
whs 已提交
3619 3620
    if dim is not None and not isinstance(dim, list):
        dim = [dim]
3621 3622 3623 3624 3625
    helper.append_op(
        type='reduce_max',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
W
whs 已提交
3626
            'dim': dim if dim != None else [0],
3627 3628 3629 3630 3631 3632
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out


C
caoying03 已提交
3633
def reduce_min(input, dim=None, keep_dim=False, name=None):
3634
    """
Y
yangyaming 已提交
3635
    Computes the minimum of tensor elements over the given dimension.
3636 3637 3638

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
W
whs 已提交
3639
        dim (list|int|None): The dimensions along which the minimum is computed.
Y
yangyaming 已提交
3640 3641 3642
            If :attr:`None`, compute the minimum over all elements of
            :attr:`input` and return a Tensor variable with a single element,
            otherwise must be in the range :math:`[-rank(input), rank(input))`.
W
whs 已提交
3643
            If :math:`dim[i] < 0`, the dimension to reduce is :math:`rank + dim[i]`.
Y
yangyaming 已提交
3644 3645
        keep_dim (bool): Whether to reserve the reduced dimension in the
            output Tensor. The result tensor will have one fewer dimension
3646
            than the :attr:`input` unless :attr:`keep_dim` is true.
C
caoying03 已提交
3647 3648
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
3649 3650 3651

    Returns:
        Variable: The reduced Tensor variable.
Y
yangyaming 已提交
3652

3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663
    Examples:
        .. code-block:: python

            # x is a Tensor variable with following elements:
            #    [[0.2, 0.3, 0.5, 0.9]
            #     [0.1, 0.2, 0.6, 0.7]]
            # Each example is followed by the correspending output tensor.
            fluid.layers.reduce_min(x)  # [0.1]
            fluid.layers.reduce_min(x, dim=0)  # [0.1, 0.2, 0.5, 0.7]
            fluid.layers.reduce_min(x, dim=-1)  # [0.2, 0.1]
            fluid.layers.reduce_min(x, dim=1, keep_dim=True)  # [[0.2], [0.1]]
W
whs 已提交
3664 3665 3666 3667 3668 3669 3670

            # x is a Tensor variable with shape [2, 2, 2] and elements as below:
            #      [[[1.0, 2.0], [3.0, 4.0]],
            #      [[5.0, 6.0], [7.0, 8.0]]]
            # Each example is followed by the correspending output tensor.
            fluid.layers.reduce_min(x, dim=[1, 2]) # [1.0, 5.0]
            fluid.layers.reduce_min(x, dim=[0, 1]) # [1.0, 2.0]
3671 3672
    """
    helper = LayerHelper('reduce_min', **locals())
X
Xin Pan 已提交
3673
    out = helper.create_variable_for_type_inference(dtype=helper.input_dtype())
W
whs 已提交
3674 3675
    if dim is not None and not isinstance(dim, list):
        dim = [dim]
3676 3677 3678 3679 3680
    helper.append_op(
        type='reduce_min',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
W
whs 已提交
3681
            'dim': dim if dim != None else [0],
3682 3683 3684 3685
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out
G
guosheng 已提交
3686 3687


3688 3689 3690 3691 3692 3693
def reduce_prod(input, dim=None, keep_dim=False, name=None):
    """
    Computes the product of tensor elements over the given dimension.

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
W
whs 已提交
3694
        dim (list|int|None): The dimensions along which the product is performed. If
3695 3696
            :attr:`None`, multipy all elements of :attr:`input` and return a
            Tensor variable with a single element, otherwise must be in the
W
whs 已提交
3697 3698
            range :math:`[-rank(input), rank(input))`. If :math:`dim[i] < 0`,
            the dimension to reduce is :math:`rank + dim[i]`.
3699 3700 3701
        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
3702
        name(str|None): A name for this layer(optional). If set None, the
Z
zhouhanqing 已提交
3703
            layer will be named automatically.
3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717

    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
3718
            fluid.layers.reduce_prod(x, dim=1,
Z
zhouhanqing 已提交
3719
                                     keep_dim=True)  # [[0.027], [0.0084]]
W
whs 已提交
3720 3721 3722 3723 3724 3725 3726

            # x is a Tensor variable with shape [2, 2, 2] and elements as below:
            #      [[[1.0, 2.0], [3.0, 4.0]],
            #      [[5.0, 6.0], [7.0, 8.0]]]
            # Each example is followed by the correspending output tensor.
            fluid.layers.reduce_prod(x, dim=[1, 2]) # [24.0, 1680.0]
            fluid.layers.reduce_prod(x, dim=[0, 1]) # [105.0, 384.0]
3727 3728
    """
    helper = LayerHelper('reduce_prod', **locals())
X
Xin Pan 已提交
3729
    out = helper.create_variable_for_type_inference(dtype=helper.input_dtype())
W
whs 已提交
3730 3731
    if dim is not None and not isinstance(dim, list):
        dim = [dim]
3732 3733 3734 3735 3736
    helper.append_op(
        type='reduce_prod',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={
W
whs 已提交
3737
            'dim': dim if dim != None else [0],
3738 3739 3740 3741 3742 3743
            'keep_dim': keep_dim,
            'reduce_all': True if dim == None else False
        })
    return out


C
caoying03 已提交
3744
def split(input, num_or_sections, dim=-1, name=None):
G
guosheng 已提交
3745
    """
C
caoying03 已提交
3746
    Split the input tensor into multiple sub-tensors.
G
guosheng 已提交
3747 3748 3749

    Args:
        input (Variable): The input variable which is a Tensor or LoDTensor.
C
caoying03 已提交
3750 3751 3752 3753 3754
        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
3755
            :attr:`dim` dimension orderly.
C
caoying03 已提交
3756
        dim (int): The dimension along which to split. If :math:`dim < 0`, the
G
guosheng 已提交
3757
            dimension to split along is :math:`rank(input) + dim`.
C
caoying03 已提交
3758 3759
        name(str|None): A name for this layer(optional). If set None, the layer
                       will be named automatically.
G
guosheng 已提交
3760 3761

    Returns:
D
dzhwinter 已提交
3762
        list(Variable): The list of segmented tensor variables.
G
guosheng 已提交
3763 3764 3765 3766 3767 3768 3769 3770 3771

    Examples:
        .. code-block:: python

            # x is a Tensor variable with shape [3, 9, 5]:
            x0, x1, x2 = fluid.layers.split(x, num_or_sections=3, dim=1)
            x0.shape  # [3, 3, 5]
            x1.shape  # [3, 3, 5]
            x2.shape  # [3, 3, 5]
F
stash  
fengjiayi 已提交
3772 3773
            x0, x1, x2 = fluid.layers.split(
                x, num_or_sections=[2, 3, 4], dim=1)
G
guosheng 已提交
3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788
            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 = [
X
Xin Pan 已提交
3789
        helper.create_variable_for_type_inference(dtype=helper.input_dtype())
G
guosheng 已提交
3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802
        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
3803 3804 3805 3806 3807 3808 3809 3810 3811


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

3812
    .. math::
3813 3814

        y = \\frac{x}{ \sqrt{\sum {x^2} + epsion }}
C
caoying03 已提交
3815 3816 3817 3818 3819

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

    Args:
3820
        x(Variable|list): The input tensor to l2_normalize layer.
3821
        axis(int): The axis on which to apply normalization. If `axis < 0`, \
3822 3823
            the dimension to normalization is rank(X) + axis. -1 is the
            last dimension.
3824
        epsilon(float): The epsilon value is used to avoid division by zero, \
3825
            the defalut value is 1e-10.
3826
        name(str|None): A name for this layer(optional). If set None, the layer \
3827
            will be named automatically.
C
caoying03 已提交
3828 3829

    Returns:
3830
        Variable: The output tensor variable is the same shape with `x`.
C
caoying03 已提交
3831 3832

    Examples:
3833

C
caoying03 已提交
3834 3835
        .. code-block:: python

3836 3837 3838 3839
            data = fluid.layers.data(name="data",
                                     shape=(3, 17, 13),
                                     dtype="float32")
            normed = fluid.layers.l2_normalize(x=data, axis=1)
C
caoying03 已提交
3840 3841
    """

F
fengjiayi 已提交
3842 3843
    if len(x.shape) == 1:
        axis = 0
C
caoying03 已提交
3844 3845
    helper = LayerHelper("l2_normalize", **locals())

X
Xin Pan 已提交
3846 3847
    out = helper.create_variable_for_type_inference(dtype=x.dtype)
    norm = helper.create_variable_for_type_inference(dtype=x.dtype)
C
caoying03 已提交
3848
    helper.append_op(
3849 3850 3851 3852
        type="norm",
        inputs={"X": x},
        outputs={"Out": out,
                 "Norm": norm},
C
caoying03 已提交
3853
        attrs={
3854 3855
            "axis": 1 if axis is None else axis,
            "epsilon": epsilon,
C
caoying03 已提交
3856 3857
        })
    return out
3858 3859


S
sneaxiy 已提交
3860
def matmul(x, y, transpose_x=False, transpose_y=False, alpha=1.0, name=None):
G
guosheng 已提交
3861
    """
Y
ying 已提交
3862 3863 3864 3865
    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
3866

C
chengduoZH 已提交
3867
    The actual behavior depends on the shapes of :math:`x`, :math:`y` and the
3868
    flag values of :attr:`transpose_x`, :attr:`transpose_y`. Specifically:
G
guosheng 已提交
3869

3870 3871 3872 3873 3874
    - 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
3875
      :math:`[1, D]` in transposed form.
G
guosheng 已提交
3876

C
chengduoZH 已提交
3877
    - After transpose, the two tensors are 2-D or n-D and matrix multiplication
3878
      performs in the following way.
G
guosheng 已提交
3879

3880
      - If both are 2-D, they are multiplied like conventional matrices.
C
chengduoZH 已提交
3881
      - If either is n-D, it is treated as a stack of matrices residing in the
Y
ying 已提交
3882
        last two dimensions and a batched matrix multiply supporting broadcast
3883
        applies on the two tensors.
G
guosheng 已提交
3884

Y
ying 已提交
3885 3886
    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
3887
    removed after matrix multiplication.
G
guosheng 已提交
3888 3889 3890

    Args:
        x (Variable): The input variable which is a Tensor or LoDTensor.
3891 3892 3893
        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.
S
sneaxiy 已提交
3894
        alpha (float): The scale of output. Default 1.0.
3895
        name(str|None): A name for this layer(optional). If set None, the layer
3896
            will be named automatically.
G
guosheng 已提交
3897 3898

    Returns:
3899
        Variable: The product Tensor variable.
G
guosheng 已提交
3900

G
guosheng 已提交
3901 3902 3903
    Examples:
        .. code-block:: python

3904
            # Examples to clarify shapes of the inputs and output
C
chengduoZH 已提交
3905 3906
            # x: [B, ..., M, K], y: [B, ..., K, N]
            fluid.layers.matmul(x, y)  # out: [B, ..., M, N]
Y
ying 已提交
3907

3908 3909
            # x: [B, M, K], y: [B, K, N]
            fluid.layers.matmul(x, y)  # out: [B, M, N]
Y
ying 已提交
3910

3911 3912
            # x: [B, M, K], y: [K, N]
            fluid.layers.matmul(x, y)  # out: [B, M, N]
Y
ying 已提交
3913

3914 3915
            # x: [M, K], y: [K, N]
            fluid.layers.matmul(x, y)  # out: [M, N]
Y
ying 已提交
3916 3917 3918 3919

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

3920 3921
            # x: [K], y: [K]
            fluid.layers.matmul(x, y)  # out: [1]
3922

Y
ying 已提交
3923
            # x: [M], y: [N]
3924
            fluid.layers.matmul(x, y, True, True)  # out: [M, N]
G
guosheng 已提交
3925
    """
Y
ying 已提交
3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937

    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
3938
            y_shape = y_shape + [1]
Y
ying 已提交
3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954

        # 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)

3955
    helper = LayerHelper('matmul', **locals())
X
Xin Pan 已提交
3956
    out = helper.create_variable_for_type_inference(dtype=x.dtype)
G
guosheng 已提交
3957
    helper.append_op(
3958 3959 3960 3961
        type='matmul',
        inputs={'X': x,
                'Y': y},
        outputs={'Out': out},
S
sneaxiy 已提交
3962 3963 3964
        attrs={
            'transpose_X': transpose_x,
            'transpose_Y': transpose_y,
S
sneaxiy 已提交
3965
            'alpha': float(alpha),
S
sneaxiy 已提交
3966
        })
3967
    return out
3968 3969


3970
def topk(input, k, name=None):
Q
qingqing01 已提交
3971 3972 3973 3974
    """
    This operator is used to find values and indices of the k largest entries
    for the last dimension.

F
fengjiayi 已提交
3975
    If the input is a vector (1-D Tensor), finds the k largest entries in the vector
Q
qingqing01 已提交
3976 3977 3978 3979 3980 3981
    and outputs their values and indices as vectors. Thus values[j] is the j-th
    largest entry in input, and its index is indices[j].

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

F
fengjiayi 已提交
3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002
    For example:

    .. code-block:: text

        If:
            input = [[5, 4, 2, 3],
                     [9, 7, 10, 25],
                     [6, 2, 10, 1]]
            k = 2

        Then:
            The first output:
            values = [[5, 4],
                      [10, 25],
                      [6, 10]]

            The second output:
            indices = [[0, 1],
                       [2, 3],
                       [0, 2]]

Q
qingqing01 已提交
4003 4004 4005
    Args:
        input(Variable): The input variable which can be a vector or Tensor with
            higher rank.
4006
        k(int):  The number of top elements to look for along the last dimension
F
fengjiayi 已提交
4007
                 of input.
4008
        name(str|None): A name for this layer(optional). If set None, the layer
4009
                       will be named automatically.
F
fengjiayi 已提交
4010
                       Default: None
Q
qingqing01 已提交
4011 4012

    Returns:
4013 4014 4015
        Tuple[Variable]: A tuple with two elements. Each element is a Variable.
        The first one is k largest elements along each last
        dimensional slice. The second one is indices of values
F
fengjiayi 已提交
4016
        within the last dimension of input.
Q
qingqing01 已提交
4017

F
fengjiayi 已提交
4018 4019
    Raises:
        ValueError: If k < 1 or k is not less than the last dimension of input
Q
qingqing01 已提交
4020 4021 4022 4023 4024 4025 4026

    Examples:
        .. code-block:: python

            top5_values, top5_indices = layers.topk(input, k=5)
    """
    helper = LayerHelper("top_k", **locals())
X
Xin Pan 已提交
4027 4028
    values = helper.create_variable_for_type_inference(dtype=input.dtype)
    indices = helper.create_variable_for_type_inference(dtype="int64")
Q
qingqing01 已提交
4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039
    helper.append_op(
        type="top_k",
        inputs={"X": [input]},
        outputs={"Out": [values],
                 "Indices": [indices]},
        attrs={"k": k})
    values.stop_gradient = True
    indices.stop_gradient = True
    return values, indices


4040
def edit_distance(input, label, normalized=True, ignored_tokens=None):
4041
    """
Y
ying 已提交
4042 4043 4044 4045 4046 4047 4048 4049 4050
    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
4051

Y
ying 已提交
4052
    "kitten" -> "sitten" -> "sittin" -> "sitting"
W
wanghaoshuang 已提交
4053

4054
    The input is a LoDTensor consisting of all the hypothesis strings with
Y
ying 已提交
4055 4056
    the total number denoted by `batch_size`, and the separation is specified
    by the LoD information. And the `batch_size` reference strings are arranged
4057
    in order in the same way in the input LoDTensor.
W
wanghaoshuang 已提交
4058

4059
    The output contains the `batch_size` results and each stands for the edit
Y
ying 已提交
4060 4061
    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
4062

4063 4064 4065
    Args:
        input(Variable): The indices for hypothesis strings.
        label(Variable): The indices for reference strings.
4066
        normalized(bool, default True): Indicated whether to normalize the edit distance by
Y
ying 已提交
4067
                          the length of reference string.
4068
        ignored_tokens(list<int>, default None): Tokens that should be removed before
Y
ying 已提交
4069
                                     calculating edit distance.
4070
        name (str): The name of this layer. It is optional.
4071

W
wanghaoshuang 已提交
4072
    Returns:
W
wanghaoshuang 已提交
4073
        Variable: sequence-to-sequence edit distance in shape [batch_size, 1].
W
wanghaoshuang 已提交
4074 4075 4076 4077

    Examples:
        .. code-block:: python

T
tink2123 已提交
4078 4079
            x = fluid.layers.data(name='x', shape=[1], dtype='float32')
            y = fluid.layers.data(name='y', shape=[1], dtype='float32')
4080
            cost = fluid.layers.edit_distance(input=x,label=y)
4081
    """
4082
    helper = LayerHelper("edit_distance", **locals())
4083

4084
    # remove some tokens from input and labels
W
wanghaoshuang 已提交
4085
    if ignored_tokens is not None and len(ignored_tokens) > 0:
X
Xin Pan 已提交
4086 4087
        erased_input = helper.create_variable_for_type_inference(dtype="int64")
        erased_label = helper.create_variable_for_type_inference(dtype="int64")
4088 4089 4090 4091 4092

        helper.append_op(
            type="sequence_erase",
            inputs={"X": [input]},
            outputs={"Out": [erased_input]},
W
wanghaoshuang 已提交
4093
            attrs={"tokens": ignored_tokens})
4094 4095 4096 4097 4098
        input = erased_input

        helper.append_op(
            type="sequence_erase",
            inputs={"X": [label]},
W
whs 已提交
4099
            outputs={"Out": [erased_label]},
W
wanghaoshuang 已提交
4100
            attrs={"tokens": ignored_tokens})
4101 4102
        label = erased_label

4103
    # edit distance op
X
Xin Pan 已提交
4104 4105
    edit_distance_out = helper.create_variable_for_type_inference(dtype="int64")
    sequence_num = helper.create_variable_for_type_inference(dtype="int64")
4106 4107 4108 4109
    helper.append_op(
        type="edit_distance",
        inputs={"Hyps": [input],
                "Refs": [label]},
4110 4111
        outputs={"Out": [edit_distance_out],
                 "SequenceNum": [sequence_num]},
4112 4113
        attrs={"normalized": normalized})

4114
    return edit_distance_out, sequence_num
4115 4116 4117 4118 4119


def ctc_greedy_decoder(input, blank, name=None):
    """
    This op is used to decode sequences by greedy policy by below steps:
Y
yi.wu 已提交
4120

Y
ying 已提交
4121 4122 4123 4124
    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.
4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141

    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]]

4142
        input.lod = [[4, 4]]
4143 4144 4145 4146 4147 4148 4149

        Then:

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

4150
        output.lod = [[2, 1]]
4151 4152 4153

    Args:

Y
ying 已提交
4154 4155 4156 4157 4158 4159 4160 4161 4162
        input(Variable): (LoDTensor<float>), the probabilities of
                         variable-length sequences, which is a 2-D Tensor with
                         LoD information. It's shape is [Lp, num_classes + 1],
                         where Lp is the sum of all input sequences' length and
                         num_classes is the true number of classes. (not
                         including the blank label).
        blank(int): the blank label index of Connectionist Temporal
                    Classification (CTC) loss, which is in thehalf-opened
                    interval [0, num_classes + 1).
4163
        name (str): The name of this layer. It is optional.
4164 4165

    Returns:
4166
        Variable: CTC greedy decode result. If all the sequences in result were
4167
        empty, the result LoDTensor will be [-1] with LoD [[]] and dims [1, 1].
4168 4169 4170 4171 4172

    Examples:
        .. code-block:: python

            x = fluid.layers.data(name='x', shape=[8], dtype='float32')
W
wanghaoshuang 已提交
4173

4174
            cost = fluid.layers.ctc_greedy_decoder(input=x, blank=0)
W
wanghaoshuang 已提交
4175
    """
4176
    helper = LayerHelper("ctc_greedy_decoder", **locals())
Q
qingqing01 已提交
4177
    _, topk_indices = topk(input, k=1)
4178 4179

    # ctc align op
X
Xin Pan 已提交
4180
    ctc_out = helper.create_variable_for_type_inference(dtype="int64")
4181 4182 4183
    helper.append_op(
        type="ctc_align",
        inputs={"Input": [topk_indices]},
W
wanghaoshuang 已提交
4184
        outputs={"Output": [ctc_out]},
4185 4186
        attrs={"merge_repeated": True,
               "blank": blank})
4187
    return ctc_out
4188 4189


W
Wu Yi 已提交
4190
def warpctc(input, label, blank=0, norm_by_times=False, use_cudnn=False):
W
wanghaoshuang 已提交
4191
    """
4192 4193
    An operator integrating the open source Warp-CTC library
    (https://github.com/baidu-research/warp-ctc)
W
wanghaoshuang 已提交
4194
    to compute Connectionist Temporal Classification (CTC) loss.
4195 4196
    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
4197 4198 4199
    input tensor.

    Args:
4200
       input (Variable): The unscaled probabilities of variable-length sequences,
W
wanghaoshuang 已提交
4201 4202 4203 4204
         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).
4205
       label (Variable): The ground truth of variable-length sequence,
4206 4207 4208
         which is a 2-D Tensor with LoD information. It is of the shape [Lg, 1],
         where Lg is th sum of all labels' length.
       blank (int, default 0): The blank label index of Connectionist
W
wanghaoshuang 已提交
4209 4210
         Temporal Classification (CTC) loss, which is in the
         half-opened interval [0, num_classes + 1).
4211 4212 4213
       norm_by_times(bool, default false): Whether to normalize the gradients
         by the number of time-step, which is also the sequence's length.
         There is no need to normalize the gradients if warpctc layer was
4214
         follewed by a mean_op.
W
Wu Yi 已提交
4215
       use_cudnn (bool, default false): Whether to use cudnn.
W
wanghaoshuang 已提交
4216 4217

    Returns:
4218 4219
        Variable: The Connectionist Temporal Classification (CTC) loss,
        which is a 2-D Tensor of the shape [batch_size, 1].
W
wanghaoshuang 已提交
4220 4221

    Examples:
4222

W
wanghaoshuang 已提交
4223
        .. code-block:: python
4224

4225 4226 4227
            label = fluid.layers.data(shape=[11, 8], dtype='float32', lod_level=1)
            predict = fluid.layers.data(shape=[11, 1], dtype='float32')
            cost = fluid.layers.warpctc(input=predict, label=label)
W
wanghaoshuang 已提交
4228 4229

    """
F
fengjiayi 已提交
4230
    helper = LayerHelper('warpctc', **locals())
X
Xin Pan 已提交
4231 4232
    loss_out = helper.create_variable_for_type_inference(dtype=input.dtype)
    grad_out = helper.create_variable_for_type_inference(dtype=input.dtype)
W
wanghaoshuang 已提交
4233 4234 4235 4236 4237 4238
    helper.append_op(
        type='warpctc',
        inputs={'Logits': [input],
                'Label': [label]},
        outputs={'WarpCTCGrad': [grad_out],
                 'Loss': [loss_out]},
W
Wu Yi 已提交
4239 4240 4241 4242 4243
        attrs={
            'blank': blank,
            'norm_by_times': norm_by_times,
            'use_cudnn': use_cudnn
        })
W
wanghaoshuang 已提交
4244
    return loss_out
4245 4246 4247 4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258 4259


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]]
4260 4261 4262
            x.data = [[1,  2], [3,  4],
                      [5,  6], [7,  8],
                      [9, 10], [11, 12]]
4263 4264 4265 4266 4267
            x.dims = [6, 2]

        set new_dim = 4

        then out is a LoDTensor:
4268

4269
            out.lod  = [[0, 1, 3]]
4270 4271 4272 4273

            out.data = [[1,  2,  3,  4],
                        [5,  6,  7,  8],
                        [9, 10, 11, 12]]
4274 4275 4276 4277 4278 4279 4280
            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:
4281 4282 4283

       input (Variable): A 2-D LoDTensor with shape being [N, M] where M for dimension.
       new_dim (int): New dimension that the input LoDTensor is reshaped to.
4284 4285

    Returns:
4286

4287 4288 4289 4290 4291
        Variable: Reshaped LoDTensor according to new dimension.

    Examples:
        .. code-block:: python

4292
            x = fluid.layers.data(shape=[5, 20], dtype='float32', lod_level=1)
4293
            x_reshaped = fluid.layers.sequence_reshape(input=x, new_dim=10)
4294 4295
    """
    helper = LayerHelper('sequence_reshape', **locals())
X
Xin Pan 已提交
4296
    out = helper.create_variable_for_type_inference(helper.input_dtype())
4297 4298 4299 4300 4301 4302
    helper.append_op(
        type='sequence_reshape',
        inputs={'X': [input]},
        outputs={'Out': [out]},
        attrs={'new_dim': new_dim})
    return out
Y
ying 已提交
4303 4304


4305 4306 4307 4308
# FIXME(wuyi): let docstring_checker.py understand @autodoc.
# For now, the comments in c++ use types like Tensor, but in python side
# the type is often "Variable", and arguments may vary.
@templatedoc(op_type="nce")
Y
Yang Yu 已提交
4309 4310 4311 4312 4313 4314
def nce(input,
        label,
        num_total_classes,
        sample_weight=None,
        param_attr=None,
        bias_attr=None,
C
chengduo 已提交
4315
        num_neg_samples=None,
4316 4317 4318 4319
        name=None,
        sampler="uniform",
        custom_dist=None,
        seed=0):
4320 4321 4322 4323 4324 4325 4326
    """
    ${comment}

    Args:
        input (Variable): input variable.
        label (Variable): label.
        num_total_classes (int):${num_total_classes_comment}
4327 4328
        sample_weight (Variable|None): A Variable of shape [batch_size, 1]
            storing a weight for each sample. The default weight for each
4329
            sample is 1.0.
C
chengduo 已提交
4330 4331 4332 4333 4334 4335 4336 4337 4338
        param_attr (ParamAttr|None): The parameter attribute for learnable parameters/weights
             of nce. If it is set to None or one attribute of ParamAttr, nce
             will create ParamAttr as param_attr. If the Initializer of the param_attr
             is not set, the parameter is initialized with Xavier. Default: None.
        bias_attr (ParamAttr|bool|None): The parameter attribute for the bias of nce.
             If it is set to False, no bias will be added to the output units.
             If it is set to None or one attribute of ParamAttr, nce
             will create ParamAttr as bias_attr. If the Initializer of the bias_attr
             is not set, the bias is initialized zero. Default: None.
4339
        num_neg_samples (int): ${num_neg_samples_comment}
C
chengduo 已提交
4340 4341
        name (str|None): A name for this layer(optional). If set None, the layer
             will be named automatically. Default: None.
4342 4343 4344 4345 4346 4347 4348 4349
        sampler (str): The sampler used to sample class from negtive classes.
                       It can be 'uniform', 'log_uniform' or 'custom_dist'.
                       default: 'uniform'.
        custom_dist (Variable): A tensor with shape [num_total_classes]. 
                       It is used when sampler is set to 'custom_dist'.
                       custom_dist[i] is the probsbility of i-th class to be sampled.
                       default: None.
        seed (int): The seed used in sampler. default: 0.
F
fengjiayi 已提交
4350

4351
    Returns:
Y
Yibing Liu 已提交
4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 4378
        Variable: The output nce loss.

    Examples:
        .. code-block:: python

            window_size = 5
            words = []
            for i in xrange(window_size):
                words.append(layers.data(
                    name='word_{0}'.format(i), shape=[1], dtype='int64'))

            dict_size = 10000
            label_word = int(window_size / 2) + 1

            embs = []
            for i in xrange(window_size):
                if i == label_word:
                    continue

                emb = layers.embedding(input=words[i], size=[dict_size, 32],
                                       param_attr='emb.w', is_sparse=True)
                embs.append(emb)

            embs = layers.concat(input=embs, axis=1)
            loss = layers.nce(input=embs, label=words[label_word],
                          num_total_classes=dict_size, param_attr='nce.w',
                          bias_attr='nce.b')
4379 4380 4381 4382 4383 4384 4385 4386 4387 4388

            #or use custom distribution
            dist = fluid.layers.assign(input=np.array([0.05,0.5,0.1,0.3,0.05]).astype("float32"))
            loss = layers.nce(input=embs, label=words[label_word],
                          num_total_classes=5, param_attr='nce.w',
                          bias_attr='nce.b',
                          num_neg_samples=3,
                          sampler="custom_dist",
                          custom_dist=dist)
            
4389
    """
Y
Yang Yu 已提交
4390 4391 4392
    helper = LayerHelper('nce', **locals())
    assert isinstance(input, Variable)
    assert isinstance(label, Variable)
C
chengduo 已提交
4393 4394

    dim = input.shape[1]
Y
Yang Yu 已提交
4395 4396 4397 4398 4399 4400
    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)
C
chengduo 已提交
4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413
    inputs = {
        'Input': input,
        'Label': label,
        'Weight': w,
        'SampleWeight': sample_weight if sample_weight is not None else []
    }
    if helper.bias_attr:
        b = helper.create_parameter(
            attr=helper.bias_attr,
            shape=[num_total_classes, 1],
            is_bias=True,
            dtype=input.dtype)
        inputs['Bias'] = b
X
Xin Pan 已提交
4414 4415 4416
    cost = helper.create_variable_for_type_inference(dtype=input.dtype)
    sample_logits = helper.create_variable_for_type_inference(dtype=input.dtype)
    sample_labels = helper.create_variable_for_type_inference(dtype=label.dtype)
Y
Yang Yu 已提交
4417

Y
Yang Yu 已提交
4418 4419 4420 4421 4422
    if num_neg_samples is None:
        num_neg_samples = 10
    else:
        num_neg_samples = int(num_neg_samples)

4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442
    inputs = {
        'Input': input,
        'Label': label,
        'Weight': w,
        'Bias': b,
        'SampleWeight': sample_weight if sample_weight is not None else []
    }

    if sampler == "uniform":
        sampler = 0
    elif sampler == "log_uniform":
        sampler = 1
    elif sampler == "custom_dist":
        assert custom_dist is not None
        assert isinstance(custom_dist, Variable)
        inputs['CustomDistribution'] = custom_dist
        sampler = 2
    else:
        raise Exception("Unsupported sampler type.")

Y
Yang Yu 已提交
4443 4444
    attrs = {
        'num_total_classes': int(num_total_classes),
4445 4446 4447
        'num_neg_samples': num_neg_samples,
        'seed': seed,
        'sampler': sampler
Y
Yang Yu 已提交
4448
    }
Y
Yang Yu 已提交
4449 4450 4451

    helper.append_op(
        type='nce',
C
chengduo 已提交
4452
        inputs=inputs,
Y
Yang Yu 已提交
4453 4454 4455 4456 4457 4458
        outputs={
            'Cost': cost,
            'SampleLogits': sample_logits,
            'SampleLabels': sample_labels
        },
        attrs=attrs)
Y
Yang Yu 已提交
4459
    return cost / (num_neg_samples + 1)
4460 4461


C
chengduo 已提交
4462 4463 4464 4465 4466 4467
def hsigmoid(input,
             label,
             num_classes,
             param_attr=None,
             bias_attr=None,
             name=None):
W
weixing02 已提交
4468 4469
    """
    The hierarchical sigmoid operator is used to accelerate the training
M
minqiyang 已提交
4470
    process of language model. This operator organizes the classes into a
G
guosheng 已提交
4471 4472 4473 4474 4475 4476 4477 4478 4479
    complete binary tree, each leaf node represents a class(a word) and each
    internal node acts as a binary classifier. For each word there's a unique
    path from root to it's leaf node, hsigmoid calculate the cost for each
    internal node on the path, and sum them to get a total cost. hsigmoid can
    achive a acceleration from :math:`O(N)` to :math:`O(logN)`, where :math:`N`
    represents the size of word dict.

    Refer to `Hierarchical Probabilistic Neural Network Language Model
    <http://www.iro.umontreal.ca/~lisa/pointeurs/hierarchical-nnlm-aistats05.pdf>`_
M
minqiyang 已提交
4480

W
weixing02 已提交
4481
    Args:
M
minqiyang 已提交
4482
        input (Variable): The input tensor variable with shape
G
guosheng 已提交
4483 4484 4485 4486 4487
            :math:`[N \\times D]`, where :math:`N` is the size of mini-batch,
            and :math:`D` is the feature size.
        label (Variable): The tensor variable contains labels of training data.
            It's a tensor with shape is :math:`[N \\times 1]`.
        num_classes: (int), The number of classes, must not be less than 2.
C
chengduo 已提交
4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498
        param_attr (ParamAttr|None): The parameter attribute for learnable parameters/weights
             of hsigmoid. If it is set to None or one attribute of ParamAttr, hsigmoid
             will create ParamAttr as param_attr. If the Initializer of the param_attr
             is not set, the parameter is initialized with Xavier. Default: None.
        bias_attr (ParamAttr|bool|None): The parameter attribute for the bias of hsigmoid.
             If it is set to False, no bias will be added to the output units.
             If it is set to None or one attribute of ParamAttr, hsigmoid
             will create ParamAttr as bias_attr. If the Initializer of the bias_attr
             is not set, the bias is initialized zero. Default: None.
        name (str|None): A name for this layer(optional). If set None, the layer
             will be named automatically. Default: None.
W
weixing02 已提交
4499 4500 4501 4502 4503 4504 4505 4506

    Returns:
        Out: (Tensor) The cost of hierarchical sigmoid operator. the shape is [N, 1]

    Examples:

        .. code-block:: python

G
guosheng 已提交
4507 4508 4509
            x = fluid.layers.data(name='x', shape=[2], dtype='float32')
            y = fluid.layers.data(name='y', shape=[1], dtype='int64')
            out = fluid.layers.hsigmoid(input=x, label=y, num_classes=6)
W
weixing02 已提交
4510 4511 4512 4513
    """

    helper = LayerHelper('hierarchical_sigmoid', **locals())
    dtype = helper.input_dtype()
X
Xin Pan 已提交
4514 4515
    out = helper.create_variable_for_type_inference(dtype)
    pre_out = helper.create_variable_for_type_inference(dtype)
W
weixing02 已提交
4516 4517
    dim = input.shape[1]
    if num_classes < 2:
G
guosheng 已提交
4518
        raise ValueError("num_classes must not be less than 2.")
W
weixing02 已提交
4519 4520 4521 4522 4523
    weights = helper.create_parameter(
        attr=helper.param_attr,
        shape=[num_classes - 1, dim],
        is_bias=False,
        dtype=input.dtype)
W
weixing02 已提交
4524 4525 4526 4527 4528 4529 4530 4531
    inputs = {"X": input, "W": weights, "Label": label}
    if helper.bias_attr:
        bias = helper.create_parameter(
            attr=helper.bias_attr,
            shape=[1, num_classes - 1],
            is_bias=True,
            dtype=input.dtype)
        inputs['Bias'] = bias
W
weixing02 已提交
4532 4533
    helper.append_op(
        type="hierarchical_sigmoid",
W
weixing02 已提交
4534
        inputs=inputs,
W
weixing02 已提交
4535 4536 4537 4538 4539 4540
        outputs={"Out": out,
                 "PreOut": pre_out},
        attrs={"num_classes": num_classes})
    return out


Y
fix ci.  
ying 已提交
4541
def transpose(x, perm, name=None):
Y
ying 已提交
4542 4543 4544 4545 4546 4547 4548
    """
    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:
4549 4550 4551
        x (Variable): The input Tensor.
        perm (list): A permutation of the dimensions of `input`.
        name (str): The name of this layer. It is optional.
Y
ying 已提交
4552 4553 4554 4555 4556 4557 4558

    Returns:
        Variable: A transposed Tensor.

    Examples:
        .. code-block:: python

4559 4560 4561 4562
            # use append_batch_size=False to avoid prepending extra 
            # batch size in shape
            x = fluid.layers.data(name='x', shape=[5, 10, 15], 
                            dtype='float32', append_batch_size=False)
Y
fix ci.  
ying 已提交
4563
            x_transposed = layers.transpose(x, perm=[1, 0, 2])
Y
ying 已提交
4564 4565
    """

Y
fix ci.  
ying 已提交
4566
    if len(perm) != len(x.shape):
Y
ying 已提交
4567 4568 4569
        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
4570 4571 4572 4573 4574 4575
    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
4576 4577

    helper = LayerHelper('transpose', **locals())
X
Xin Pan 已提交
4578 4579
    out = helper.create_variable_for_type_inference(x.dtype)
    x_shape = helper.create_variable_for_type_inference(x.dtype)
Y
ying 已提交
4580
    helper.append_op(
4581
        type='transpose2',
Y
fix ci.  
ying 已提交
4582
        inputs={'X': [x]},
4583 4584
        outputs={'Out': [out],
                 'XShape': [x_shape]},
Y
ying 已提交
4585 4586
        attrs={'axis': perm})
    return out
4587 4588


4589 4590 4591 4592 4593 4594 4595
def im2sequence(input,
                filter_size=1,
                stride=1,
                padding=0,
                input_image_size=None,
                out_stride=1,
                name=None):
4596
    """
4597 4598 4599 4600 4601 4602 4603
    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:
4604 4605 4606 4607 4608 4609 4610 4611 4612 4613

    .. 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
4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628 4629 4630 4631

        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.

4632 4633 4634 4635 4636 4637 4638 4639 4640
        input_image_size(Variable): the input contains image real size.It's dim
            is [batchsize, 2]. It is dispensable.It is just for batch inference.

        out_stride(int|tuple): The scaling of image through CNN. It is
            dispensable. It is valid only when input_image_size is not null.
            If out_stride is tuple,  it must contain two intergers,
            (out_stride_H, out_stride_W). Otherwise,
            the out_stride_H = out_stride_W = out_stride.

4641 4642 4643
        name (int): The name of this layer. It is optional.

    Returns:
W
wanghaoshuang 已提交
4644 4645 4646 4647 4648
        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.
4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675

    Examples:

        .. code-block:: text

            Given:

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

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

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

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

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

            And:

W
wanghaoshuang 已提交
4676 4677 4678
            filter = [2, 2]
            stride = [1, 1]
            padding = [0, 0]
4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690

            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.]]

4691
            output.dims = {8, 8}
4692

4693
            output.lod = [[4, 4]]
4694

D
dzhwinter 已提交
4695
     Examples:
4696 4697 4698

        .. code-block:: python

4699 4700
            output = fluid.layers.im2sequence(
                input=layer, stride=[1, 1], filter_size=[2, 2])
4701 4702

    """
W
wanghaoshuang 已提交
4703 4704 4705 4706 4707 4708 4709 4710 4711 4712

    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])
4713 4714 4715 4716 4717 4718 4719
    inputs = {"X": input}
    attrs = {"kernels": filter_size, "strides": stride, "padding": padding}
    if input_image_size:
        if isinstance(out_stride, int):
            out_stride = [out_stride, out_stride]
        inputs["Y"] = input_image_size
        attrs["out_stride"] = out_stride
4720
    helper = LayerHelper('im2sequence', **locals())
X
Xin Pan 已提交
4721
    out = helper.create_variable_for_type_inference(dtype=helper.input_dtype())
4722
    helper.append_op(
4723
        type='im2sequence', inputs=inputs, outputs={'Out': out}, attrs=attrs)
4724
    return out
4725 4726


Y
yuyang18 已提交
4727
@templatedoc()
4728
def row_conv(input, future_context_size, param_attr=None, act=None):
Y
yuyang18 已提交
4729 4730
    """
    ${comment}
4731 4732

    Args:
Y
yuyang18 已提交
4733
        input (${x_type}): ${x_comment}.
Y
yangyaming 已提交
4734 4735
        future_context_size (int): Future context size. Please note, the shape
            of convolution kernel is [future_context_size + 1, D].
4736 4737 4738 4739 4740
        param_attr (ParamAttr): Attributes of parameters, including
            name, initializer etc.
        act (str): Non-linear activation to be applied to output variable.

    Returns:
Y
yuyang18 已提交
4741
        ${out_comment}.
4742 4743

    Examples:
Y
yuyang18 已提交
4744 4745 4746 4747
        >>> import paddle.fluid as fluid
        >>> x = fluid.layers.data(name='x', shape=[16],
        >>>                        dtype='float32', lod_level=1)
        >>> out = fluid.layers.row_conv(input=x, future_context_size=2)
4748 4749 4750 4751 4752 4753
    """
    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)
X
Xin Pan 已提交
4754
    out = helper.create_variable_for_type_inference(dtype)
4755 4756 4757 4758 4759
    helper.append_op(
        type='row_conv',
        inputs={'X': [input],
                'Filter': [filter_param]},
        outputs={'Out': [out]})
Y
yangyaming 已提交
4760
    return helper.append_activation(out)
4761 4762


Y
yuyang18 已提交
4763
@templatedoc()
4764 4765
def multiplex(inputs, index):
    """
Y
yuyang18 已提交
4766 4767 4768 4769 4770 4771 4772
    ${comment}

    >>> import paddle.fluid as fluid
    >>> 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)
4773 4774

    Args:
Y
yuyang18 已提交
4775 4776
       inputs (list): ${x_comment}.
       index (${ids_type}): ${ids_comment}.
4777 4778

    Returns:
Y
yuyang18 已提交
4779
        ${out_comment}.
4780 4781
    """
    helper = LayerHelper('multiplex', **locals())
Y
yangyaming 已提交
4782 4783 4784 4785 4786

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

X
Xin Pan 已提交
4787
    out = helper.create_variable_for_type_inference(inputs[0].dtype)
4788 4789 4790 4791 4792 4793
    helper.append_op(
        type='multiplex',
        inputs={'X': inputs,
                'Ids': index},
        outputs={'Out': [out]})
    return out
4794 4795


4796 4797 4798
def softmax_with_cross_entropy(logits,
                               label,
                               soft_label=False,
S
sneaxiy 已提交
4799
                               ignore_index=-100,
4800 4801
                               numeric_stable_mode=False,
                               return_softmax=False):
4802 4803
    """
    **Softmax With Cross Entropy Operator.**
4804

4805 4806 4807 4808
    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.
4809

4810 4811 4812
    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.
4813

4814 4815 4816
    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.
4817

4818
    The equation is as follows:
4819

4820
    1) Hard label (one-hot label, so every sample has exactly one class)
4821

4822 4823 4824 4825
    .. math::

        loss_j =  -\\text{logit}_{label_j} +
        \\log\\left(\\sum_{i=0}^{K}\\exp(\\text{logit}_i)\\right), j = 1,..., K
4826

4827 4828 4829
    2) Soft label (each sample can have a distribution over all classes)

    .. math::
4830

4831 4832 4833 4834
        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

S
sneaxiy 已提交
4835 4836 4837 4838 4839 4840 4841 4842 4843 4844 4845 4846
    3) If numeric_stable_mode is True, softmax is calculated first by:

    .. math::
        
        max_j = \\max_{i=0}^{K}{\\text{logit}_i}

        log\\_max\\_sum_j = \\log\\sum_{i=0}^{K}\\exp(logit_i - max_j)

        softmax_j = \\exp(logit_j - max_j - {log\\_max\\_sum}_j)

    and then cross entropy loss is calculated by softmax and label.

4847 4848 4849 4850 4851 4852 4853 4854
    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.
M
minqiyang 已提交
4855 4856
        ignore_index (int): Specifies a target value that is ignored and does
                            not contribute to the input gradient. Only valid
4857
                            if soft_label is set to False. Default: -100
S
sneaxiy 已提交
4858 4859 4860 4861 4862 4863 4864
        numeric_stable_mode (bool): A flag to indicate whether to use a more
                                    numerically stable algorithm. Only valid
                                    when soft_label is False and GPU is used.
                                    When soft_label is True or CPU is used, 
                                    the algorithm is always numerically stable. 
                                    Note that the speed may be slower when use 
                                    stable algorithm. Default: False
4865 4866
        return_softmax (bool): A flag indicating whether to return the softmax 
                               along with the cross entropy loss. Default: False
4867

4868
    Returns:
4869 4870 4871 4872 4873
        Variable or Tuple of two Variables: Return the cross entropy loss if 
                              `return_softmax` is False, otherwise the tuple 
                              (loss, softmax), where the cross entropy loss is 
                              a 2-D tensor with shape [N x 1], and softmax is a 
                              2-D tensor with shape [N x K].
4874 4875 4876 4877 4878 4879 4880

    Examples:
        .. code-block:: python

            data = fluid.layers.data(name='data', shape=[128], dtype='float32')
            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
            fc = fluid.layers.fc(input=data, size=100)
F
stash  
fengjiayi 已提交
4881 4882
            out = fluid.layers.softmax_with_cross_entropy(
                logits=fc, label=label)
4883 4884
    """
    helper = LayerHelper('softmax_with_cross_entropy', **locals())
X
Xin Pan 已提交
4885 4886
    softmax = helper.create_variable_for_type_inference(dtype=logits.dtype)
    loss = helper.create_variable_for_type_inference(dtype=logits.dtype)
4887 4888 4889 4890 4891 4892
    helper.append_op(
        type='softmax_with_cross_entropy',
        inputs={'Logits': logits,
                'Label': label},
        outputs={'Softmax': softmax,
                 'Loss': loss},
S
sneaxiy 已提交
4893 4894 4895 4896 4897
        attrs={
            'soft_label': soft_label,
            'ignore_index': ignore_index,
            'numeric_stable_mode': numeric_stable_mode
        })
4898 4899 4900 4901

    if return_softmax:
        return loss, softmax

4902 4903 4904 4905 4906
    return loss


def smooth_l1(x, y, inside_weight=None, outside_weight=None, sigma=None):
    """
Y
Yibing Liu 已提交
4907 4908
    This layer computes the smooth L1 loss for Variable :attr:`x` and :attr:`y`.
    It takes the first dimension of :attr:`x` and :attr:`y` as batch size.
Q
qingqing01 已提交
4909
    For each instance, it computes the smooth L1 loss element by element first
4910
    and then sums all the losses. So the shape of ouput Variable is
4911
    [batch_size, 1].
4912

4913 4914
    Args:
        x (Variable): A tensor with rank at least 2. The input value of smooth
Q
qingqing01 已提交
4915
            L1 loss op with shape [batch_size, dim1, ..., dimN].
4916
        y (Variable): A tensor with rank at least 2. The target value of smooth
Y
Yibing Liu 已提交
4917
            L1 loss op with same shape as :attr:`x`.
4918
        inside_weight (Variable|None):  A tensor with rank at least 2. This
4919 4920
            input is optional and should have same shape with :attr:`x`. If
            provided, the result of (:attr:`x` - :attr:`y`) will be multiplied
Y
Yibing Liu 已提交
4921
            by this tensor element by element.
4922
        outside_weight (Variable|None): A tensor with rank at least 2. This
4923 4924
            input is optional and should have same shape with :attr:`x`. If
            provided, the out smooth L1 loss will be multiplied by this tensor
Y
Yibing Liu 已提交
4925
            element by element.
4926
        sigma (float|None): Hyper parameter of smooth L1 loss layer. A float
4927 4928
           scalar with default value 1.0.

4929
    Returns:
4930
        Variable: The output smooth L1 loss with shape [batch_size, 1].
4931 4932 4933 4934 4935

    Examples:
        .. code-block:: python

            data = fluid.layers.data(name='data', shape=[128], dtype='float32')
F
stash  
fengjiayi 已提交
4936 4937
            label = fluid.layers.data(
                name='label', shape=[100], dtype='float32')
4938
            fc = fluid.layers.fc(input=data, size=100)
F
fengjiayi 已提交
4939
            out = fluid.layers.smooth_l1(x=fc, y=label)
4940
    """
4941

4942
    helper = LayerHelper('smooth_l1_loss', **locals())
X
Xin Pan 已提交
4943 4944
    diff = helper.create_variable_for_type_inference(dtype=x.dtype)
    loss = helper.create_variable_for_type_inference(dtype=x.dtype)
4945 4946 4947 4948 4949 4950 4951 4952 4953 4954 4955 4956
    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
4957 4958 4959 4960


def one_hot(input, depth):
    """
Y
Yibing Liu 已提交
4961
    This layer creates the one-hot representations for input indices.
4962 4963

    Args:
Y
Yibing Liu 已提交
4964 4965
        input(Variable): Input indices, last dimension must be 1.
        depth(scalar): An interger defining the depth of the one-hot dimension.
4966 4967

    Returns:
Y
Yibing Liu 已提交
4968
        Variable: The one-hot representations of input.
4969 4970

    Examples:
C
caoying03 已提交
4971
        .. code-block:: python
4972

Y
Yibing Liu 已提交
4973 4974
            label = layers.data(name="label", shape=[1], dtype="float32")
            one_hot_label = layers.one_hot(input=label, depth=10)
4975 4976
    """
    helper = LayerHelper("one_hot", **locals())
X
Xin Pan 已提交
4977
    one_hot_out = helper.create_variable_for_type_inference(dtype='float32')
4978 4979 4980 4981 4982 4983
    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
4984 4985


Y
Yu Yang 已提交
4986
def autoincreased_step_counter(counter_name=None, begin=1, step=1):
Y
Yu Yang 已提交
4987
    """
Y
yi.wu 已提交
4988 4989 4990
    Create an auto-increase variable
    which will be automatically increased by 1 every mini-batch
    Return the run counter of the main program, default is started from 1.
Y
Yu Yang 已提交
4991 4992 4993 4994 4995 4996

    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.

4997 4998
    Returns:
        Variable: The global run counter.
Y
yi.wu 已提交
4999 5000 5001 5002 5003 5004

    Examples:
        .. code-block:: python

           global_step = fluid.layers.autoincreased_step_counter(
               counter_name='@LR_DECAY_COUNTER@', begin=begin, step=1)
Y
Yu Yang 已提交
5005 5006
    """
    helper = LayerHelper('global_step_counter')
Y
Yu Yang 已提交
5007 5008
    if counter_name is None:
        counter_name = '@STEP_COUNTER@'
Y
Yu Yang 已提交
5009 5010 5011 5012 5013
    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
5014
                value=begin - 1, force_cpu=True))
W
Wu Yi 已提交
5015
        helper.main_program.global_block()._prepend_op(
Y
Yu Yang 已提交
5016 5017
            type='increment',
            inputs={'X': [counter]},
Y
Yu Yang 已提交
5018 5019
            outputs={'Out': [counter]},
            attrs={'step': float(step)})
Y
Yu Yang 已提交
5020 5021 5022
        counter.stop_gradient = True

    return counter
Y
yangyaming 已提交
5023 5024


5025
def reshape(x, shape, actual_shape=None, act=None, inplace=False, name=None):
C
caoying03 已提交
5026
    """
C
caoying03 已提交
5027 5028
    Gives a new shape to the input Tensor without changing its data.

5029 5030 5031 5032 5033
    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
5034

5035
    Some tricks exist when specifying the target shape.
C
caoying03 已提交
5036

5037 5038 5039 5040
    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.

5041
    2. 0 means the actual dimension value is going to be copied from the
5042 5043 5044 5045
    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
5046 5047

    1. Given a 3-D tensor x with a shape [2, 4, 6], and the target shape
W
wanghaoshuang 已提交
5048
    is [6, 8], the reshape operator will transform x into a 2-D tensor with
5049
    shape [6, 8] and leaving x's data unchanged.
C
caoying03 已提交
5050

5051
    2. Given a 3-D tensor x with a shape [2, 4, 6], and the target shape
5052 5053
    specified is [2, 3, -1, 2], the reshape operator will transform x into a
    4-D tensor with shape [2, 3, 4, 2] and leaving x's data unchanged. In this
W
wanghaoshuang 已提交
5054 5055
    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
5056
    dimensions.
C
caoying03 已提交
5057

5058
    3. Given a 3-D tensor x with a shape [2, 4, 6], and the target shape
5059 5060 5061 5062
    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
5063 5064

    Args:
5065
        x(variable): The input tensor.
C
caoying03 已提交
5066 5067
        shape(list): The new shape. At most one dimension of the new shape can
                     be -1.
5068 5069 5070 5071 5072
        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`.
5073 5074
        act (str): The non-linear activation to be applied to the reshaped tensor
                   variable.
5075 5076 5077 5078 5079 5080 5081
        inplace(bool): Must use :attr:`False` if :attr:`x` is used in multiple
                       operators. If this flag is set :attr:`True`, reuse input
                       :attr:`x` to reshape, which will change the shape of
                       tensor variable :attr:`x` and might cause errors when
                       :attr:`x` is used in multiple operators. If :attr:`False`,
                       preserve the shape :attr:`x` and create a new output tensor
                       variable whose data is copied from input x but reshaped.
5082
        name (str): The name of this layer. It is optional.
C
caoying03 已提交
5083

5084
    Returns:
G
guosheng 已提交
5085 5086 5087 5088
        Variable: The reshaped tensor variable if :attr:`act` is None. It is a \
                  new tensor variable if :attr:`inplace` is :attr:`False`, \
                  otherwise it is :attr:`x`. If :attr:`act` is not None, return \
                  the activated tensor variable.
C
caoying03 已提交
5089

X
Xin Pan 已提交
5090 5091 5092
    Raises:
        TypeError: if actual_shape is neither Variable nor None.

C
caoying03 已提交
5093 5094
    Examples:
        .. code-block:: python
G
guosheng 已提交
5095

5096
            data = fluid.layers.data(
5097
                name='data', shape=[2, 4, 6], dtype='float32')
C
caoying03 已提交
5098
            reshaped = fluid.layers.reshape(
G
guosheng 已提交
5099
                x=data, shape=[-1, 0, 3, 2], inplace=True)
C
caoying03 已提交
5100 5101 5102
    """

    if not (isinstance(shape, list) or isinstance(shape, tuple)):
L
luotao1 已提交
5103
        raise ValueError("Input shape must be a python list or tuple.")
X
Xin Pan 已提交
5104 5105 5106 5107 5108
    inputs = {"X": x}
    if isinstance(actual_shape, Variable):
        inputs["Shape"] = actual_shape
    elif actual_shape is not None:
        raise TypeError("actual_shape should either be Variable or None")
C
caoying03 已提交
5109

5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124
    # 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.")

5125
    helper = LayerHelper("reshape2", **locals())
5126 5127
    out = x if inplace else helper.create_variable_for_type_inference(
        dtype=x.dtype)
X
Xin Pan 已提交
5128
    x_shape = helper.create_variable_for_type_inference(dtype=x.dtype)
C
caoying03 已提交
5129
    helper.append_op(
5130
        type="reshape2",
X
Xin Pan 已提交
5131
        inputs=inputs,
D
dzhwinter 已提交
5132
        attrs={"shape": shape},
5133 5134
        outputs={"Out": out,
                 "XShape": x_shape})
C
caoying03 已提交
5135

D
dzhwinter 已提交
5136
    return helper.append_activation(out)
5137

5138

5139
def squeeze(input, axes, name=None):
Y
Yibing Liu 已提交
5140
    """
M
minqiyang 已提交
5141 5142 5143
    Remove single-dimensional entries from the shape of a tensor. Takes a
    parameter axes with a list of axes to squeeze. If axes is not provided, all
    the single dimensions will be removed from the shape. If an axis is
Y
Yibing Liu 已提交
5144
    selected with shape entry not equal to one, an error is raised.
M
minqiyang 已提交
5145

Y
Yibing Liu 已提交
5146 5147
    Examples:
    Case 1:
M
minqiyang 已提交
5148
      Given
Y
Yibing Liu 已提交
5149 5150 5151 5152 5153 5154 5155 5156
        X.shape = (1, 3, 1, 5)
      and
        axes = [0]
      we get:
        Out.shape = (3, 1, 5)
      Case 2:
        Given
          X.shape = (1, 3, 1, 5)
M
minqiyang 已提交
5157
        and
Y
Yibing Liu 已提交
5158 5159 5160
          axes = []
        we get:
          Out.shape = (3, 5)
M
minqiyang 已提交
5161

Y
Yibing Liu 已提交
5162
    Args:
5163
        input (Variable): The input variable to be squeezed.
Y
Yibing Liu 已提交
5164
        axes (list): List of integers, indicating the dimensions to be squeezed.
5165
        name (str|None): Name for this layer.
Y
Yibing Liu 已提交
5166 5167 5168 5169 5170 5171 5172 5173

    Returns:
        Variable: Output squeezed variable.

    Examples:
        .. code-block:: python

            x = layers.data(name='x', shape=[5, 1, 10])
5174
            y = layers.sequeeze(input=x, axes=[1])
Y
Yibing Liu 已提交
5175 5176
    """
    helper = LayerHelper("squeeze", **locals())
X
Xin Pan 已提交
5177 5178
    out = helper.create_variable_for_type_inference(dtype=input.dtype)
    x_shape = helper.create_variable_for_type_inference(dtype=input.dtype)
Y
Yibing Liu 已提交
5179
    helper.append_op(
5180
        type="squeeze2",
5181
        inputs={"X": input},
Y
Yibing Liu 已提交
5182
        attrs={"axes": axes},
5183 5184
        outputs={"Out": out,
                 "XShape": x_shape})
Y
Yibing Liu 已提交
5185

5186 5187 5188
    return out


5189
def unsqueeze(input, axes, name=None):
Y
Yibing Liu 已提交
5190
    """
M
minqiyang 已提交
5191 5192 5193
    Insert single-dimensional entries to the shape of a tensor. Takes one
    required argument axes, a list of dimensions that will be inserted.
    Dimension indices in axes are as seen in the output tensor.
Y
Yibing Liu 已提交
5194

M
minqiyang 已提交
5195 5196
    For example:
      Given a tensor such that tensor with shape [3, 4, 5],
Y
Yibing Liu 已提交
5197
      then Unsqueezed tensor with axes=[0, 4] has shape [1, 3, 4, 5, 1].
M
minqiyang 已提交
5198

Y
Yibing Liu 已提交
5199
    Args:
5200
        input (Variable): The input variable to be unsqueezed.
Y
Yibing Liu 已提交
5201
        axes (list): List of integers, indicating the dimensions to be inserted.
5202
        name (str|None): Name for this layer.
Y
Yibing Liu 已提交
5203 5204 5205 5206 5207 5208 5209 5210

    Returns:
        Variable: Output unsqueezed variable.

    Examples:
        .. code-block:: python

            x = layers.data(name='x', shape=[5, 10])
5211
            y = layers.unsequeeze(input=x, axes=[1])
Y
Yibing Liu 已提交
5212 5213
    """
    helper = LayerHelper("unsqueeze", **locals())
X
Xin Pan 已提交
5214 5215
    out = helper.create_variable_for_type_inference(dtype=input.dtype)
    x_shape = helper.create_variable_for_type_inference(dtype=input.dtype)
Y
Yibing Liu 已提交
5216
    helper.append_op(
5217
        type="unsqueeze2",
5218
        inputs={"X": input},
Y
Yibing Liu 已提交
5219
        attrs={"axes": axes},
5220 5221
        outputs={"Out": out,
                 "XShape": x_shape})
Y
Yibing Liu 已提交
5222

5223 5224
    return out

5225

Y
yangyaming 已提交
5226
def lod_reset(x, y=None, target_lod=None):
Y
yangyaming 已提交
5227
    """
Y
Yibing Liu 已提交
5228
    Set LoD of :attr:`x` to a new one specified by :attr:`y` or
5229 5230 5231 5232
    :attr:`target_lod`. When :attr:`y` provided, :attr:`y.lod` would be
    considered as target LoD first, otherwise :attr:`y.data` would be
    considered as target LoD. If :attr:`y` is not provided, target LoD should
    be specified by :attr:`target_lod`. If target LoD is specified by
Y
Yibing Liu 已提交
5233
    :attr:`Y.data` or :attr:`target_lod`, only one level LoD is supported.
Y
yangyaming 已提交
5234 5235 5236 5237 5238 5239

    .. code-block:: text

        * Example 1:

            Given a 1-level LoDTensor x:
5240
                x.lod =  [[ 2,           3,                   1 ]]
Y
yangyaming 已提交
5241 5242 5243
                x.data = [[1.0], [2.0], [3.0], [4.0], [5.0], [6.0]]
                x.dims = [6, 1]

5244
            target_lod: [4, 2]
Y
yangyaming 已提交
5245 5246

            then we get a 1-level LoDTensor:
5247
                out.lod =  [[4,                          2]]
Y
yangyaming 已提交
5248 5249 5250 5251 5252 5253
                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:
5254
                x.lod =  [[2,            3,                   1]]
Y
yangyaming 已提交
5255 5256 5257 5258
                x.data = [[1.0], [2.0], [3.0], [4.0], [5.0], [6.0]]
                x.dims = [6, 1]

            y is a Tensor:
5259
                y.data = [[2, 4]]
Y
yangyaming 已提交
5260 5261 5262
                y.dims = [1, 3]

            then we get a 1-level LoDTensor:
5263
                out.lod =  [[2,            4]]
Y
yangyaming 已提交
5264 5265 5266 5267 5268 5269
                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:
5270
                x.lod =  [[2,            3,                   1]]
Y
yangyaming 已提交
5271 5272 5273 5274
                x.data = [[1.0], [2.0], [3.0], [4.0], [5.0], [6.0]]
                x.dims = [6, 1]

            y is a 2-level LoDTensor:
5275
                y.lod =  [[2, 2], [2, 2, 1, 1]]
Y
yangyaming 已提交
5276 5277 5278 5279
                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:
5280
                out.lod =  [[2, 2], [2, 2, 1, 1]]
Y
yangyaming 已提交
5281 5282 5283 5284 5285
                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.
5286
        y (Variable|None): If provided, output's LoD would be derived
Y
Yibing Liu 已提交
5287
                           from :attr:`y`.
Y
yangyaming 已提交
5288
        target_lod (list|tuple|None): One level LoD which should be considered
Y
Yibing Liu 已提交
5289
                                      as target LoD when :attr:`y` not provided.
Y
yangyaming 已提交
5290 5291

    Returns:
Y
Yibing Liu 已提交
5292
        Variable: Output variable with LoD specified by this layer.
Y
yangyaming 已提交
5293 5294

    Raises:
Y
Yibing Liu 已提交
5295
        ValueError: If :attr:`y` and :attr:`target_lod` are both None.
Y
yangyaming 已提交
5296 5297 5298 5299 5300 5301 5302 5303 5304

    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())
X
Xin Pan 已提交
5305
    out = helper.create_variable_for_type_inference(dtype=x.dtype)
Y
yangyaming 已提交
5306 5307 5308 5309 5310 5311 5312 5313 5314 5315 5316 5317 5318 5319
    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
5320 5321 5322 5323 5324 5325 5326 5327 5328 5329 5330


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

    The formula is as follows:

    .. math::

D
dzhwinter 已提交
5331
      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}
D
dragonwarrior 已提交
5332 5333 5334 5335 5336 5337 5338 5339 5340 5341 5342 5343 5344 5345 5346 5347 5348 5349 5350 5351 5352 5353 5354 5355 5356 5357 5358 5359

    In the above equation:

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

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

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

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

    Returns:
        A tensor variable storing the transformation result.

    Examples:
        .. code-block:: python

F
stash  
fengjiayi 已提交
5360 5361
          data = fluid.layers.data(
              name="data", shape=[3, 112, 112], dtype="float32")
D
dragonwarrior 已提交
5362 5363 5364 5365 5366 5367 5368 5369 5370 5371 5372 5373
          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))

X
Xin Pan 已提交
5374 5375 5376
    mid_out = helper.create_variable_for_type_inference(
        dtype=dtype, stop_gradient=True)
    lrn_out = helper.create_variable_for_type_inference(dtype)
D
dragonwarrior 已提交
5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389
    helper.append_op(
        type="lrn",
        inputs={"X": input},
        outputs={
            "Out": lrn_out,
            "MidOut": mid_out,
        },
        attrs={"n": n,
               "k": k,
               "alpha": alpha,
               "beta": beta})

    return lrn_out
G
guosheng 已提交
5390 5391 5392 5393


def pad(x, paddings, pad_value=0., name=None):
    """
G
guosheng 已提交
5394
    Pads a tensor with a constant value given by :attr:`pad_value`, and the
W
wanghaoshuang 已提交
5395
    padded width is specified by :attr:`paddings`.
G
guosheng 已提交
5396

G
guosheng 已提交
5397 5398 5399 5400
    Specifically, the number of values padded before the contents of :attr:`x`
    in dimension :attr:`i` is indicated by :attr:`paddings[i]`, and the number
    of values padded after the contents of :attr:`x` in dimension :attr:`i` is
    indicated by :attr:`paddings[i+1]`.
G
guosheng 已提交
5401 5402 5403 5404 5405 5406 5407 5408 5409 5410 5411 5412 5413 5414 5415 5416 5417 5418 5419 5420 5421 5422

    See below for an example.

    .. code-block:: text

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

            paddings = [0, 1, 1, 2]

            pad_value = 0

        Return:

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

    Args:
        x (Variable): The input tensor variable.
        paddings (list): A list of integers. Its elements specify the padded
                         width before and after for each dimension in turn.
W
wanghaoshuang 已提交
5423
                         The length of :attr:paddings must be
G
guosheng 已提交
5424 5425 5426 5427 5428 5429 5430 5431 5432 5433
                         :math:`rank(x) \\times 2`.
        pad_value (float): The constant value used to pad.
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.

    Returns:
        Variable: The padded tensor variable.

    Examples:
        .. code-block:: python
G
guosheng 已提交
5434

G
guosheng 已提交
5435 5436 5437 5438 5439 5440
            # x is a rank 2 tensor variable.
            out = fluid.layers.pad(
                x=x, paddings=[0, 1, 1, 2], pad_value=0.)
    """
    helper = LayerHelper('pad', input=x, **locals())
    dtype = helper.input_dtype()
X
Xin Pan 已提交
5441
    out = helper.create_variable_for_type_inference(dtype)
G
guosheng 已提交
5442 5443 5444 5445 5446 5447 5448
    helper.append_op(
        type='pad',
        inputs={'X': x},
        outputs={'Out': out},
        attrs={'paddings': paddings,
               'pad_value': float(pad_value)})
    return out
5449 5450


C
chengduo 已提交
5451 5452 5453 5454 5455 5456 5457 5458 5459 5460 5461 5462 5463 5464 5465 5466 5467 5468 5469 5470 5471 5472 5473 5474 5475 5476 5477 5478 5479 5480 5481 5482 5483 5484 5485 5486 5487 5488 5489 5490 5491 5492 5493 5494 5495 5496 5497 5498 5499 5500 5501 5502 5503 5504 5505 5506 5507 5508 5509 5510 5511 5512 5513 5514 5515 5516 5517 5518 5519 5520
def pad_constant_like(x, y, pad_value=0., name=None):
    """
    Pad input(Y) with :attr:`pad_value`, the number of values padded to
    the edges of each axis is specified by the difference of the shape
    of X and Y. ((0, shape_x_0 - shape_y_0), ... (0, shape_x_n - shape_y_n))
    unique pad widths for each axis. The input should be a k-D
    tensor(k > 0 and k < 7).

    See below for an example.

    .. code-block:: text

        Given:
            X = [[[[ 0,  1,  2],
                   [ 3,  4,  5]],
                  [[ 6,  7,  8],
                   [ 9, 10, 11]],
                  [[12, 13, 14],
                   [15, 16, 17]]],
                 [[[18, 19, 20],
                   [21, 22, 23]],
                  [[24, 25, 26],
                   [27, 28, 29]],
                  [[30, 31, 32],
                   [33, 34, 35]]]]
            X.shape = (2, 3, 2, 3)

            Y = [[[[35, 36, 37]],
                  [[38, 39, 40]],
                  [[41, 42, 43]]]]
            Y.shape = (1, 3, 1, 3)

    And
        pad_value = -1,

    Return:
        Out = [[[[35, 36, 37],
                  [-1, -1, -1]],
                [[38, 39, 40],
                  [-1, -1, -1]],
                 [[41, 42, 43],
                  [-1, -1, -1]]],
                [[[-1, -1, -1],
                  [-1, -1, -1]],
                 [[-1, -1, -1],
                  [-1, -1, -1]],
                 [[-1, -1, -1],
                  [-1, -1, -1]]]]
        Out.shape = (2, 3, 2, 3)

    Args:
        x (Variable): The input tensor variable.
        y (Variable): The input tensor variable.
        pad_value (float): The constant value used to pad.
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.

    Returns:
        Variable: The padded tensor variable.

    Examples:
        .. code-block:: python

            # x is a rank 4 tensor variable, x.shape = (2, 3, 2, 3)
            # y is a rank 4 tensor variable, y.shape = (1, 3, 1, 3)
            out = fluid.layers.pad_constant_like(x=x, y=y, pad_value=0.)
            # out is a rank 4 tensor variable, and out.shape = [2, 3 ,2 , 3]
    """
    helper = LayerHelper('pad_constant_like', input=x, **locals())
    dtype = helper.input_dtype()
X
Xin Pan 已提交
5521
    out = helper.create_variable_for_type_inference(dtype)
C
chengduo 已提交
5522 5523 5524 5525 5526 5527 5528 5529 5530
    helper.append_op(
        type='pad_constant_like',
        inputs={'X': x,
                'Y': y},
        outputs={'Out': out},
        attrs={'pad_value': float(pad_value)})
    return out


5531 5532 5533 5534 5535 5536 5537
def label_smooth(label,
                 prior_dist=None,
                 epsilon=0.1,
                 dtype="float32",
                 name=None):
    """
    Label smoothing is a mechanism to regularize the classifier layer and is
5538 5539
    called label-smoothing regularization (LSR).

5540 5541 5542 5543 5544 5545 5546 5547 5548 5549 5550 5551 5552 5553 5554 5555 5556 5557 5558 5559 5560 5561 5562
    Label smoothing is proposed to encourage the model to be less confident,
    since optimizing the log-likelihood of the correct label directly may
    cause overfitting and reduce the ability of the model to adapt. Label
    smoothing replaces the ground-truth label :math:`y` with the weighted sum
    of itself and some fixed distribution :math:`\mu`. For class :math:`k`,
    i.e.

    .. math::

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

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

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

    Args:
        label(Variable): The input variable containing the label data. The
                          label data should use one-hot representation.
        prior_dist(Variable): The prior distribution to be used to smooth
                              labels. If not provided, an uniform distribution
                              is used. The shape of :attr:`prior_dist` should
5563
                              be :math:`(1, class\_num)`.
5564 5565
        epsilon(float): The weight used to mix up the original ground-truth
                        distribution and the fixed distribution.
5566
        dtype(np.dtype|core.VarDesc.VarType|str): The type of data : float32,
5567 5568 5569 5570 5571 5572 5573 5574 5575 5576 5577 5578 5579 5580 5581 5582 5583 5584 5585
                                                  float_64, int etc.
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.

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

    Examples:
        .. code-block:: python

            label = layers.data(name="label", shape=[1], dtype="float32")
            one_hot_label = layers.one_hot(input=label, depth=10)
            smooth_label = layers.label_smooth(
                label=one_hot_label, epsilon=0.1, dtype="float32")
    """
    if epsilon > 1. or epsilon < 0.:
        raise ValueError("The value of epsilon must be between 0 and 1.")
    helper = LayerHelper("label_smooth", **locals())
    label.stop_gradient = True
X
Xin Pan 已提交
5586
    smooth_label = helper.create_variable_for_type_inference(dtype)
5587 5588 5589 5590 5591 5592 5593
    helper.append_op(
        type="label_smooth",
        inputs={"X": label,
                "PriorDist": prior_dist} if prior_dist else {"X": label},
        outputs={"Out": smooth_label},
        attrs={"epsilon": float(epsilon)})
    return smooth_label
5594 5595


P
peizhilin 已提交
5596
if os.name != 'nt':
P
peizhilin 已提交
5597

P
peizhilin 已提交
5598
    @templatedoc()
P
peizhilin 已提交
5599 5600 5601 5602 5603
    def roi_pool(input,
                 rois,
                 pooled_height=1,
                 pooled_width=1,
                 spatial_scale=1.0):
P
peizhilin 已提交
5604 5605
        """
        ${comment}
5606

P
peizhilin 已提交
5607 5608 5609 5610 5611 5612
        Args:
            input (Variable): ${x_comment}
            rois (Variable): ROIs (Regions of Interest) to pool over.
            pooled_height (integer): ${pooled_height_comment} Default: 1
            pooled_width (integer): ${pooled_width_comment} Default: 1
            spatial_scale (float): ${spatial_scale_comment} Default: 1.0
5613

P
peizhilin 已提交
5614 5615
        Returns:
            Variable: ${out_comment}.
5616

P
peizhilin 已提交
5617 5618
        Examples:
            .. code-block:: python
5619

P
peizhilin 已提交
5620 5621 5622 5623 5624 5625 5626 5627 5628 5629 5630 5631 5632 5633 5634 5635 5636 5637
                pool_out = fluid.layers.roi_pool(input=x, rois=rois, 7, 7, 1.0)
        """
        helper = LayerHelper('roi_pool', **locals())
        dtype = helper.input_dtype()
        pool_out = helper.create_variable_for_type_inference(dtype)
        argmaxes = helper.create_variable_for_type_inference(dtype='int32')
        helper.append_op(
            type="roi_pool",
            inputs={"X": input,
                    "ROIs": rois},
            outputs={"Out": pool_out,
                     "Argmax": argmaxes},
            attrs={
                "pooled_height": pooled_height,
                "pooled_width": pooled_width,
                "spatial_scale": spatial_scale
            })
        return pool_out
W
whs 已提交
5638 5639


J
jerrywgz 已提交
5640 5641 5642 5643 5644 5645
@templatedoc()
def roi_align(input,
              rois,
              pooled_height=1,
              pooled_width=1,
              spatial_scale=1.0,
J
jerrywgz 已提交
5646 5647
              sampling_ratio=-1,
              name=None):
J
jerrywgz 已提交
5648 5649 5650 5651 5652 5653 5654 5655 5656 5657 5658 5659 5660 5661 5662 5663
    """
    ${comment}

    Args:
        input (Variable): ${x_comment}
        rois (Variable): ROIs (Regions of Interest) to pool over.
        pooled_height (integer): ${pooled_height_comment} Default: 1
        pooled_width (integer): ${pooled_width_comment} Default: 1
        spatial_scale (float): ${spatial_scale_comment} Default: 1.0
        sampling_ratio(intger): ${sampling_ratio_comment} Default: -1

    Returns:
        Variable: ${out_comment}.
    Examples:
        .. code-block:: python

5664 5665 5666
            align_out = fluid.layers.roi_align(input=x,
                                               rois=rois,
                                               pooled_height=7,
J
jerrywgz 已提交
5667 5668 5669 5670 5671 5672
                                               pooled_width=7,
                                               spatial_scale=0.5,
                                               sampling_ratio=-1)
    """
    helper = LayerHelper('roi_align', **locals())
    dtype = helper.input_dtype()
X
Xin Pan 已提交
5673
    align_out = helper.create_variable_for_type_inference(dtype)
J
jerrywgz 已提交
5674 5675 5676 5677 5678 5679 5680 5681 5682 5683 5684 5685 5686 5687
    helper.append_op(
        type="roi_align",
        inputs={"X": input,
                "ROIs": rois},
        outputs={"Out": align_out},
        attrs={
            "pooled_height": pooled_height,
            "pooled_width": pooled_width,
            "spatial_scale": spatial_scale,
            "sampling_ratio": sampling_ratio
        })
    return align_out


W
whs 已提交
5688 5689 5690 5691 5692 5693 5694 5695 5696 5697 5698 5699 5700 5701 5702 5703 5704 5705 5706 5707 5708 5709 5710 5711 5712 5713
def dice_loss(input, label, epsilon=0.00001):
    """
    Dice loss for comparing the similarity of two batch of data,
    usually is used for binary image segmentation i.e. labels are binary.
    The dice loss can be defined as below equation:

    .. math::

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


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

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

    Examples:
5714 5715
        .. code-block:: python

W
whs 已提交
5716 5717 5718 5719
            predictions = fluid.layers.softmax(x)
            loss = fluid.layers.dice_loss(input=predictions, label=label, 2)
    """
    label = one_hot(label, depth=input.shape[-1])
5720
    reduce_dim = list(range(1, len(input.shape)))
W
whs 已提交
5721 5722 5723 5724 5725 5726
    inse = reduce_sum(input * label, dim=reduce_dim)
    dice_denominator = reduce_sum(
        input, dim=reduce_dim) + reduce_sum(
            label, dim=reduce_dim)
    dice_score = 1 - inse * 2 / (dice_denominator + epsilon)
    return reduce_mean(dice_score)
5727 5728


5729 5730 5731 5732
def image_resize(input,
                 out_shape=None,
                 scale=None,
                 name=None,
5733 5734
                 resample='BILINEAR',
                 actual_shape=None):
5735
    """
Q
qiaolongfei 已提交
5736
    **Resize a Batch of Images**
F
stash  
fengjiayi 已提交
5737

5738
    The input must be a tensor of the shape (num_batches, channels, in_h, in_w),
5739 5740 5741
    and the resizing only applies on the last two dimensions(hight and width).

    Supporting resample methods:
Q
update  
qiaolongfei 已提交
5742

5743
        'BILINEAR' : Bilinear interpolation
5744
        'NEAREST' : Nearest neighbor interpolation
F
stash  
fengjiayi 已提交
5745

5746
    Args:
5747
        input (Variable): The input tensor of image resize layer,
5748 5749
                          This is a 4-D tensor of the shape
                          (num_batches, channels, in_h, in_w).
5750
        out_shape(list|tuple|Variable|None): Output shape of image resize
5751 5752
                                    layer, the shape is (out_h, out_w).
                                    Default: None
B
baiyf 已提交
5753
        scale(float|None): The multiplier for the input height or width.
5754 5755 5756
                         At least one of out_shape or scale must be set.
                         And out_shape has a higher priority than scale.
                         Default: None
5757 5758
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.
5759 5760
        resample(str): The resample method. It supports 'BILINEAR' and 'NEAREST' 
                       currently.
5761
                       Default: 'BILINEAR'
5762 5763 5764 5765 5766 5767 5768 5769 5770 5771 5772 5773 5774
        actual_shape(Variable): An optional input to specify output shape 
                                dynamically. If provided, image resize  
                                according to this given shape rather than 
                                :attr:`out_shape` and :attr:`scale` specifying
                                shape. That is to say actual_shape has the 
                                highest priority. It is recommended to use 
                                actual_shape instead of :attr:`out_shape` if you 
                                want to specify output shape dynamically. When 
                                using actual_shape to specify output shape, one of 
                                :attr:`out_shape` and :attr:`scale` should also be 
                                set, otherwise errors would be occured in graph 
                                constructing stage.
                                Default: None
5775 5776

    Returns:
Q
update  
qiaolongfei 已提交
5777 5778
        Variable: The output is a 4-D tensor of the shape
        (num_batches, channls, out_h, out_w).
F
stash  
fengjiayi 已提交
5779

5780 5781 5782 5783 5784 5785 5786 5787
    Raises:
        TypeError: out_shape should be a list or tuple or Variable.
        TypeError: actual_shape should either be Variable or None.
        ValueError: The 'resample' of image_resize can only be 'BILINEAR' 
                    or 'NEAREST' currently.
        ValueError: One of out_shape and scale must not be None.
        ValueError: out_shape length should be 2.

5788 5789 5790
    Examples:
        .. code-block:: python

5791
            out = fluid.layers.image_resize(input, out_shape=[12, 12], resample="NEAREST")
5792
    """
5793 5794 5795 5796
    resample_methods = {
        'BILINEAR': 'bilinear',
        'NEAREST': 'nearest',
    }
5797 5798
    if resample not in resample_methods:
        raise ValueError(
5799
            "The 'resample' of image_resize can only be 'BILINEAR' or 'NEAREST' currently."
5800
        )
5801
    if out_shape is None and scale is None:
5802
        raise ValueError("One of out_shape and scale must not be None.")
5803
    helper = LayerHelper('interpolate', **locals())
5804
    dtype = helper.input_dtype()
5805 5806 5807 5808

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

5809 5810 5811
    out_h = 0
    out_w = 0
    inputs = {"X": input}
5812
    if out_shape is not None:
5813 5814 5815 5816
        if isinstance(out_shape, Variable):
            warnings.warn("out_shape as Variable type is deprecated, \
                    it is recommended to use actual_shape instead of \
                    out_shape to specify output shape dynamically.")
5817
            inputs['OutSize'] = out_shape
5818 5819 5820 5821 5822 5823 5824 5825
        elif not (_is_list_or_turple_(out_shape)):
            raise TypeError("out_shape should be a list or tuple or Variable.")
        elif len(out_shape) != 2:
            raise ValueError("out_shape length should be 2.")

        out_shape = list(map(int, out_shape))
        out_h = out_shape[0]
        out_w = out_shape[1]
5826 5827 5828 5829
    else:
        out_h = int(input.shape[2] * scale)
        out_w = int(input.shape[3] * scale)

5830 5831 5832 5833 5834
    if isinstance(actual_shape, Variable):
        inputs["OutSize"] = actual_shape
    elif actual_shape is not None:
        raise TypeError("actual_shape should either be Variable or None.")

X
Xin Pan 已提交
5835
    out = helper.create_variable_for_type_inference(dtype)
5836
    helper.append_op(
5837
        type='interpolate',
5838
        inputs=inputs,
5839
        outputs={"Out": out},
5840 5841 5842 5843 5844
        attrs={
            "out_h": out_h,
            "out_w": out_w,
            "interp_method": resample_methods[resample]
        })
5845
    return out
F
stash  
fengjiayi 已提交
5846 5847


5848
@templatedoc(op_type="interpolate")
5849 5850 5851 5852 5853
def resize_bilinear(input,
                    out_shape=None,
                    scale=None,
                    name=None,
                    actual_shape=None):
5854
    """
5855 5856 5857 5858 5859 5860 5861 5862 5863 5864 5865 5866
    Resize input by performing bilinear interpolation based on given 
    output shape which specified by actual_shape, out_shape and scale 
    in priority order.

    Bilinear interpolation is an extension of linear interpolation for 
    interpolating functions of two variables (e.g. H-direction and 
    W-direction in this op) on a rectilinear 2D grid. The key idea is 
    to perform linear interpolation first in one direction, and then 
    again in the other direction.

    For details of bilinear interpolation, please refer to Wikipedia: 
    https://en.wikipedia.org/wiki/Bilinear_interpolation
Y
yuyang18 已提交
5867 5868 5869 5870 5871

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

        out_shape(${out_size_type}): ${out_size_comment}.
5872

Y
yuyang18 已提交
5873 5874 5875 5876 5877
        scale(float|None): The multiplier for the input height or width. At
             least one of out_shape or scale must be set. And out_shape has
             a higher priority than scale. Default: None.

        name(str|None): The output variable name.
5878 5879 5880 5881 5882 5883 5884 5885 5886 5887 5888 5889 5890
        actual_shape(Variable): An optional input to specify output shape 
                                dynamically. If provided, image resize  
                                according to this given shape rather than 
                                :attr:`out_shape` and :attr:`scale` specifying
                                shape. That is to say actual_shape has the 
                                highest priority. It is recommended to use 
                                actual_shape instead of :attr:`out_shape` if you 
                                want to specify output shape dynamically. When 
                                using actual_shape to specify output shape, one of 
                                :attr:`out_shape` and :attr:`scale` should also be 
                                set, otherwise errors would be occured in graph 
                                constructing stage.
                                Default: None
Y
yuyang18 已提交
5891 5892 5893

    Returns:
        ${out_comment}.
5894 5895 5896 5897 5898

    Examples:
        .. code-block:: python

            out = fluid.layers.resize_bilinear(input, out_shape=[12, 12])
5899 5900
    """

5901
    return image_resize(input, out_shape, scale, name, 'BILINEAR', actual_shape)
5902 5903


5904
@templatedoc(op_type="interpolate")
5905 5906 5907 5908 5909
def resize_nearest(input,
                   out_shape=None,
                   scale=None,
                   name=None,
                   actual_shape=None):
5910
    """
5911 5912 5913 5914 5915 5916 5917
    Resize input by performing nearest neighbor interpolation in both the
    3rd dimention(in height direction) and the 4th dimention(in width 
    direction) based on given output shape which specified by actual_shape, 
    out_shape and scale in priority order.

    For details of nearest neighbor interpolation, please refer to Wikipedia: 
    https://en.wikipedia.org/wiki/Nearest-neighbor_interpolation
Y
yuyang18 已提交
5918 5919 5920 5921 5922

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

        out_shape(${out_size_type}): ${out_size_comment}.
5923

Y
yuyang18 已提交
5924 5925 5926 5927 5928
        scale(float|None): The multiplier for the input height or width. At
             least one of out_shape or scale must be set. And out_shape has
             a higher priority than scale. Default: None.

        name(str|None): The output variable name.
5929 5930 5931 5932 5933 5934 5935 5936 5937 5938 5939 5940 5941
        actual_shape(Variable): An optional input to specify output shape 
                                dynamically. If provided, image resize  
                                according to this given shape rather than 
                                :attr:`out_shape` and :attr:`scale` specifying
                                shape. That is to say actual_shape has the 
                                highest priority. It is recommended to use 
                                actual_shape instead of :attr:`out_shape` if you 
                                want to specify output shape dynamically. When 
                                using actual_shape to specify output shape, one of 
                                :attr:`out_shape` and :attr:`scale` should also be 
                                set, otherwise errors would be occured in graph 
                                constructing stage.
                                Default: None
Y
yuyang18 已提交
5942 5943 5944

    Returns:
        ${out_comment}.
5945 5946 5947 5948 5949

    Examples:
        .. code-block:: python

            out = fluid.layers.resize_nearest(input, out_shape=[12, 12])
5950 5951
    """

5952
    return image_resize(input, out_shape, scale, name, 'NEAREST', actual_shape)
5953 5954 5955 5956


def image_resize_short(input, out_short_len, resample='BILINEAR'):
    """
5957 5958 5959
    Resize a batch of images. The short edge of input images will be
    resized to the given 'out_short_len'. The long edge of input images
    will be resized proportionately to make images' length-width ratio
5960 5961 5962 5963 5964 5965 5966
    constant.

    Args:
        input (Variable): The input tensor of image resize layer,
                          This is a 4-D tensor of the shape
                          (num_batches, channels, in_h, in_w).
        out_short_len(int): The length of output images' short edge.
5967
        resample (str): resample method, default: BILINEAR.
F
fengjiayi 已提交
5968

5969
    Returns:
Q
update  
qiaolongfei 已提交
5970
        Variable: The output is a 4-D tensor of the shape
5971
        (num_batches, channls, out_h, out_w).
5972 5973 5974 5975 5976 5977 5978 5979 5980 5981
    """
    in_shape = input.shape
    if len(in_shape) != 4:
        raise ValueError(
            "The rank of input must be 4 (num_batches, channels, in_h, in_w).")
    hw = in_shape[2:4]
    short_idx = hw.index(min(hw))
    long_idx = 1 - short_idx
    out_shape = list(hw)
    out_shape[short_idx] = out_short_len
F
fengjiayi 已提交
5982 5983 5984
    out_shape[long_idx] = int(
        float(out_shape[long_idx]) * (float(out_short_len) / float(hw[
            short_idx])) + 0.5)
5985 5986 5987
    return image_resize(input=input, out_shape=out_shape, resample=resample)


W
whs 已提交
5988 5989
def gather(input, index):
    """
Q
qiaolongfei 已提交
5990 5991
    **Gather Layer**

5992
    Output is obtained by gathering entries of the outer-most dimension
W
whs 已提交
5993 5994 5995 5996
    of X indexed by `index` and concatenate them together.

    .. math::

5997
        Out = X[Index]
W
whs 已提交
5998 5999 6000 6001 6002 6003 6004


    .. code-block:: text


                Given:

6005 6006
                X = [[1, 2],
                     [3, 4],
W
whs 已提交
6007 6008 6009 6010 6011 6012 6013 6014 6015 6016
                     [5, 6]]

                Index = [1, 2]

                Then:

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

    Args:
6017
        input (Variable): The source input with rank>=1.
W
whs 已提交
6018 6019 6020 6021 6022 6023
        index (Variable): The index input with rank=1.

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

    Examples:
W
whs 已提交
6024

W
whs 已提交
6025 6026 6027 6028 6029 6030
        .. code-block:: python

            output = fluid.layers.gather(x, index)
    """
    helper = LayerHelper('gather', **locals())
    dtype = helper.input_dtype()
X
Xin Pan 已提交
6031
    out = helper.create_variable_for_type_inference(dtype)
W
whs 已提交
6032 6033 6034 6035 6036 6037 6038 6039
    helper.append_op(
        type="gather",
        inputs={"X": input,
                "Index": index},
        outputs={"Out": out})
    return out


6040 6041 6042 6043 6044 6045 6046 6047 6048 6049 6050 6051 6052 6053 6054 6055 6056 6057 6058 6059 6060 6061 6062 6063 6064 6065 6066 6067 6068 6069 6070
def scatter(input, index, updates, name=None):
    """
    **Scatter Layer**

    Output is obtained by updating the input on selected indices on the first
    axis.

    .. math::

        Out = X
        Out[Ids] = Updates

    Args:
        input (Variable): The source input with rank>=1.
        index (Variable): The index input with rank=1. Its dtype should be
                          int32 or int64 as it is used as indexes.
        updates (Variable): The updated value of scatter op.
        name (str|None): The output variable name. Default None.

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

    Examples:

        .. code-block:: python

            output = fluid.layers.scatter(input, index, updates)

    """
    helper = LayerHelper('scatter', **locals())
    dtype = helper.input_dtype()
X
Xin Pan 已提交
6071
    out = helper.create_variable_for_type_inference(dtype)
6072 6073 6074 6075 6076 6077 6078 6079 6080
    helper.append_op(
        type="scatter",
        inputs={"X": input,
                "Ids": index,
                "Updates": updates},
        outputs={"Out": out})
    return out


Q
Qingsheng Li 已提交
6081 6082 6083 6084 6085 6086 6087 6088 6089 6090 6091 6092 6093 6094 6095 6096 6097 6098 6099 6100 6101 6102 6103 6104 6105 6106 6107 6108 6109 6110 6111 6112 6113 6114 6115 6116 6117 6118 6119 6120 6121 6122 6123 6124 6125 6126 6127 6128 6129 6130
def sequence_scatter(input, index, updates, name=None):
    """
    **Sequence Scatter Layer**

    This operator scatters the Updates tensor to the input X. It uses the LoD
    information of Ids to select the rows to update, and use the values in Ids as
    the columns to update in each row of X.

    Here is an example:
    Given the following input:
    .. code-block:: text
        input.data = [[1.0, 1.0, 1.0, 1.0, 1.0, 1.0],
                      [1.0, 1.0, 1.0, 1.0, 1.0, 1.0],
                      [1.0, 1.0, 1.0, 1.0, 1.0, 1.0]]
        input.dims = [3, 6]

        index.data = [[0], [1], [2], [5], [4], [3], [2], [1], [3], [2], [5], [4]]
        index.lod =  [[0,        3,                       8,                 12]]

        updates.data = [[0.3], [0.3], [0.4], [0.1], [0.2], [0.3], [0.4], [0.0], [0.2], [0.3], [0.1], [0.4]]
        updates.lod =  [[  0,            3,                                 8,                         12]]

    Then we have the output:
    .. code-block:: text
        out.data = [[1.3, 1.3, 1.4, 1.0, 1.0, 1.0],
                    [1.0, 1.0, 1.4, 1.3, 1.2, 1.1],
                    [1.0, 1.0, 1.3, 1.2, 1.4, 1.1]]
        out.dims = X.dims = [3, 6]

    Args:
        input (Variable): The source input with rank>=1.
        index (Variable): A LoD Tensor. The index input of sequence scatter op
            where input will be  updated. The index input with rank=1. Its dtype
            should be int32 or int64 as it is used as indexes.
        updates (Variable): A LoD Tensor. The values to scatter to the input
            tensor X, must be a LoDTensor with the same LoD information as index.
        name (str|None): The output variable name. Default None.

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

    Examples:

        .. code-block:: python

            output = fluid.layers.sequence_scatter(input, index, updates)

    """
    helper = LayerHelper('sequence_scatter', **locals())
    dtype = helper.input_dtype()
X
Xin Pan 已提交
6131
    out = helper.create_variable_for_type_inference(dtype)
Q
Qingsheng Li 已提交
6132 6133 6134 6135 6136 6137 6138 6139 6140
    helper.append_op(
        type="sequence_scatter",
        inputs={"X": input,
                "Ids": index,
                "Updates": updates},
        outputs={"Out": out})
    return out


Y
yuyang18 已提交
6141 6142 6143 6144 6145 6146 6147 6148 6149 6150 6151 6152 6153
@templatedoc()
def random_crop(x, shape, seed=None):
    """
    ${comment}

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

    Returns:
        ${out_comment}
6154

6155 6156 6157
    Examples:
        >>> img = fluid.layers.data("img", [3, 256, 256])
        >>> cropped_img = fluid.layers.random_crop(img, shape=[3, 224, 224])
Y
yuyang18 已提交
6158
    """
F
stash  
fengjiayi 已提交
6159
    helper = LayerHelper("random_crop", **locals())
F
fengjiayi 已提交
6160
    dtype = x.dtype
X
Xin Pan 已提交
6161
    out = helper.create_variable_for_type_inference(dtype)
Y
yuyang18 已提交
6162
    if seed is None:
6163
        seed = np.random.randint(-65536, 65536)
F
fengjiayi 已提交
6164
    op_attrs = {"shape": shape}
F
stash  
fengjiayi 已提交
6165
    if isinstance(seed, int):
F
fengjiayi 已提交
6166 6167 6168 6169 6170
        op_attrs["startup_seed"] = seed
        seed = helper.create_variable(
            name=unique_name.generate("random_crop_seed"),
            dtype="int64",
            persistable=True)
F
stash  
fengjiayi 已提交
6171 6172 6173 6174
    elif not isinstance(seed, Variable):
        raise ValueError("'seed' must be a Variable or an int.")
    helper.append_op(
        type="random_crop",
F
fix  
fengjiayi 已提交
6175
        inputs={"X": x,
F
stash  
fengjiayi 已提交
6176 6177
                "Seed": seed},
        outputs={"Out": out,
F
fengjiayi 已提交
6178 6179
                 "SeedOut": seed},
        attrs=op_attrs)
F
stash  
fengjiayi 已提交
6180
    return out
W
whs 已提交
6181 6182


6183
def log(x, name=None):
W
wanghaoshuang 已提交
6184 6185 6186 6187 6188
    """
    Calculates the natural log of the given input tensor, element-wise.

    .. math::

6189
        Out = \\ln(x)
W
wanghaoshuang 已提交
6190 6191

    Args:
6192
        x (Variable): Input tensor.
6193 6194
        name (str|None, default None): A name for this layer If set None,
            the layer will be named automatically.
W
wanghaoshuang 已提交
6195 6196 6197 6198 6199 6200 6201 6202

    Returns:
        Variable: The natural log of the input tensor computed element-wise.

    Examples:

        .. code-block:: python

6203
            output = fluid.layers.log(x)
W
wanghaoshuang 已提交
6204 6205
    """
    helper = LayerHelper('log', **locals())
W
wanghaoshuang 已提交
6206
    dtype = helper.input_dtype(input_param_name='x')
X
Xin Pan 已提交
6207
    out = helper.create_variable_for_type_inference(dtype)
W
wanghaoshuang 已提交
6208
    helper.append_op(type="log", inputs={"X": x}, outputs={"Out": out})
W
wanghaoshuang 已提交
6209 6210 6211
    return out


6212
def relu(x, name=None):
W
wanghaoshuang 已提交
6213 6214
    """
    Relu takes one input data (Tensor) and produces one output data (Tensor)
6215
    where the rectified linear function, y = max(0, x), is applied to
W
wanghaoshuang 已提交
6216 6217 6218 6219
    the tensor elementwise.

    .. math::

6220
        Out = \\max(0, x)
W
wanghaoshuang 已提交
6221 6222

    Args:
6223
        x (Variable): The input tensor.
6224 6225
        name (str|None, default None): A name for this layer If set None,
            the layer will be named automatically.
W
wanghaoshuang 已提交
6226 6227 6228 6229 6230 6231 6232 6233

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

    Examples:

        .. code-block:: python

6234
            output = fluid.layers.relu(x)
W
wanghaoshuang 已提交
6235 6236
    """
    helper = LayerHelper('relu', **locals())
W
wanghaoshuang 已提交
6237
    dtype = helper.input_dtype(input_param_name='x')
X
Xin Pan 已提交
6238
    out = helper.create_variable_for_type_inference(dtype)
W
wanghaoshuang 已提交
6239
    helper.append_op(type="relu", inputs={"X": x}, outputs={"Out": out})
W
wanghaoshuang 已提交
6240
    return out
6241 6242


C
chengduo 已提交
6243 6244 6245 6246 6247 6248 6249 6250 6251 6252 6253 6254 6255 6256 6257 6258 6259 6260 6261 6262 6263 6264 6265 6266 6267 6268 6269 6270 6271 6272 6273 6274 6275 6276 6277 6278 6279 6280 6281 6282 6283
@templatedoc()
def selu(x, scale=None, alpha=None, name=None):
    """
    ${comment}

    Args:
        x (Variable): The input tensor.
        scale(float, None): If the scale is not set,
            the default value is 1.0507009873554804934193349852946.
            For more information about this value, please refer
            to: https://arxiv.org/abs/1706.02515.
        alpha(float, None): If the alpha is not set,
            the default value is 1.6732632423543772848170429916717.
            For more information about this value, please refer
            to: https://arxiv.org/abs/1706.02515.
        name (str|None, default None): A name for this layer If set None,
            the layer will be named automatically.

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

    Examples:

        .. code-block:: python

            output = fluid.layers.selu(x)
    """
    helper = LayerHelper('selu', **locals())
    dtype = helper.input_dtype(input_param_name='x')
    out = helper.create_variable_for_type_inference(dtype)
    attrs = {}
    if scale is not None:
        attrs["scale"] = scale
    if alpha is not None:
        attrs["alpha"] = alpha

    helper.append_op(
        type="selu", inputs={"X": x}, outputs={"Out": out}, attrs=attrs)
    return out


W
whs 已提交
6284 6285 6286
def mean_iou(input, label, num_classes):
    """
    Mean Intersection-Over-Union is a common evaluation metric for
6287 6288 6289 6290
    semantic image segmentation, which first computes the IOU for each
    semantic class and then computes the average over classes.
    IOU is defined as follows:

W
whs 已提交
6291
    .. math::
6292 6293

        IOU = \\frac{true\_positiv}{(true\_positive + false\_positive + false\_negative)}.
W
whs 已提交
6294

6295
    The predictions are accumulated in a confusion matrix and mean-IOU
W
whs 已提交
6296 6297 6298 6299 6300
    is then calculated from it.


    Args:
        input (Variable): A Tensor of prediction results for semantic labels with type int32 or int64.
6301
        label (Variable): A Tensor of ground truth labels with type int32 or int64.
W
whs 已提交
6302
                           Its shape should be the same as input.
6303
        num_classes (int): The possible number of labels.
W
whs 已提交
6304 6305 6306 6307

    Returns:
        mean_iou (Variable): A Tensor representing the mean intersection-over-union with shape [1].
        out_wrong(Variable): A Tensor with shape [num_classes]. The wrong numbers of each class.
6308
        out_correct(Variable): A Tensor with shape [num_classes]. The correct numbers of each class.
W
whs 已提交
6309 6310 6311 6312

    Examples:

        .. code-block:: python
6313

W
whs 已提交
6314 6315 6316 6317
            iou, wrongs, corrects = fluid.layers.mean_iou(predict, label, num_classes)
    """
    helper = LayerHelper('mean_iou', **locals())
    dtype = helper.input_dtype()
X
Xin Pan 已提交
6318 6319 6320
    out_mean_iou = helper.create_variable_for_type_inference(dtype='float32')
    out_wrong = helper.create_variable_for_type_inference(dtype='int32')
    out_correct = helper.create_variable_for_type_inference(dtype='int32')
W
whs 已提交
6321 6322
    helper.append_op(
        type="mean_iou",
W
whs 已提交
6323 6324
        inputs={"Predictions": input,
                "Labels": label},
W
whs 已提交
6325
        outputs={
W
whs 已提交
6326 6327 6328
            "OutMeanIou": out_mean_iou,
            "OutWrong": out_wrong,
            "OutCorrect": out_correct
W
whs 已提交
6329 6330 6331
        },
        attrs={"num_classes": num_classes})
    return out_mean_iou, out_wrong, out_correct
6332 6333 6334 6335 6336 6337 6338 6339 6340 6341 6342 6343 6344 6345 6346 6347 6348 6349 6350 6351 6352 6353 6354 6355 6356 6357 6358 6359 6360 6361 6362 6363 6364 6365 6366 6367 6368 6369 6370 6371 6372 6373 6374 6375 6376 6377 6378 6379 6380 6381 6382 6383 6384 6385 6386 6387 6388 6389 6390 6391 6392 6393 6394 6395 6396 6397 6398 6399 6400 6401 6402 6403 6404 6405


def crop(x, shape=None, offsets=None, name=None):
    """
    Crop input into output, as specified by offsets and shape.

    .. code-block:: text

        * Case 1:
            Given
                X = [[0, 1, 2, 0, 0]
                     [0, 3, 4, 0, 0]
                     [0, 0, 0, 0, 0]],
            and
                shape = [2, 2],
                offsets = [0, 1],
            output is:
                Out = [[1, 2],
                       [3, 4]].
        * Case 2:
            Given
                X = [[0, 1, 2, 5, 0]
                     [0, 3, 4, 6, 0]
                     [0, 0, 0, 0, 0]],
            and shape is tensor
                shape = [[0, 0, 0]
                         [0, 0, 0]]
            and
                offsets = [0, 1],

            output is:
                Out = [[1, 2, 5],
                       [3, 4, 6]].

    Args:
        x (Variable): The input tensor variable.
        shape (Variable|list/tuple of integer): The output shape is specified
            by `shape`, which can a Variable or a list/tupe of integer.
            If a tensor Variable, it's rank must be the same as `x`. This way
            is suitable for the case that the output shape may be changed each
            iteration. If a list/tupe of integer, it's length must be the same
            as the rank of `x`
        offsets (Variable|list/tuple of integer|None): Specifies the copping
            offsets at each dimension. It can be a Variable or or a list/tupe
            of integer. If a tensor Variable, it's rank must be the same as `x`.
            This way is suitable for the case that the offsets may be changed
            each iteration. If a list/tupe of integer, it's length must be the
            same as the rank of `x`. If None, the offsets are 0 at each
            dimension.
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.

    Returns:
        Variable: The cropped tensor variable.

    Raises:
        ValueError: If shape is not a list, tuple or Variable.

    Examples:

        .. code-block:: python

            x = fluid.layers.data(name="x", shape=[3, 5], dtype="float32")
            y = fluid.layers.data(name="y", shape=[2, 3], dtype="float32")
            crop = fluid.layers.crop(x, shape=y)

            # or
            z = fluid.layers.data(name="z", shape=[3, 5], dtype="float32")
            crop = fluid.layers.crop(z, shape=[2, 3])

    """
    helper = LayerHelper('crop', **locals())

    if not (isinstance(shape, list) or isinstance(shape, tuple) or \
C
chengduo 已提交
6406
                    isinstance(shape, Variable)):
6407 6408 6409 6410 6411
        raise ValueError("The shape should be a list, tuple or Variable.")

    if offsets is None:
        offsets = [0] * len(x.shape)

X
Xin Pan 已提交
6412
    out = helper.create_variable_for_type_inference(x.dtype)
6413 6414 6415 6416 6417 6418 6419 6420 6421 6422 6423 6424 6425 6426 6427 6428 6429
    ipts = {'X': x}
    attrs = {}
    if isinstance(shape, Variable):
        ipts['Y'] = shape
    else:
        attrs['shape'] = shape
    if isinstance(offsets, Variable):
        ipts['Offsets'] = offsets
    else:
        attrs['offsets'] = offsets

    helper.append_op(
        type='crop',
        inputs=ipts,
        outputs={'Out': out},
        attrs=None if len(attrs) == 0 else attrs)
    return out
6430 6431


W
whs 已提交
6432 6433 6434 6435 6436 6437 6438 6439 6440 6441 6442 6443 6444 6445 6446 6447 6448 6449 6450 6451 6452 6453 6454 6455 6456 6457 6458 6459 6460 6461 6462 6463 6464 6465 6466 6467 6468 6469 6470 6471 6472 6473 6474 6475 6476 6477 6478 6479 6480 6481 6482 6483 6484 6485 6486 6487 6488 6489 6490 6491 6492 6493 6494 6495 6496 6497 6498 6499 6500 6501 6502 6503 6504 6505 6506 6507 6508 6509 6510 6511 6512 6513 6514 6515 6516 6517 6518 6519 6520 6521 6522 6523 6524 6525 6526 6527 6528 6529 6530 6531 6532 6533 6534 6535 6536 6537 6538 6539 6540 6541 6542 6543 6544 6545 6546 6547 6548 6549
def affine_grid(theta, out_shape, name=None):
    """
    It generates a grid of (x,y) coordinates using the parameters of
    the affine transformation that correspond to a set of points where
    the input feature map should be sampled to produce the transformed
    output feature map.

    .. code-block:: text

        * Case 1:

          Given:

              theta = [[[x_11, x_12, x_13]
                        [x_14, x_15, x_16]]
                       [[x_21, x_22, x_23]
                        [x_24, x_25, x_26]]]
      
              out_shape = [2, 3, 5, 5]
      
          Step 1:
      
              Generate normalized coordinates according to out_shape.
              The values of the normalized coordinates are in the interval between -1 and 1.
              The shape of the normalized coordinates is [2, H, W] as below:
      
              C = [[[-1.  -1.  -1.  -1.  -1. ]
                    [-0.5 -0.5 -0.5 -0.5 -0.5]
                    [ 0.   0.   0.   0.   0. ]
                    [ 0.5  0.5  0.5  0.5  0.5]
                    [ 1.   1.   1.   1.   1. ]]
                   [[-1.  -0.5  0.   0.5  1. ]
                    [-1.  -0.5  0.   0.5  1. ]
                    [-1.  -0.5  0.   0.5  1. ]
                    [-1.  -0.5  0.   0.5  1. ]
                    [-1.  -0.5  0.   0.5  1. ]]]
              C[0] is the coordinates in height axis and  C[1] is the coordinates in width axis.

          Step2:

              Tanspose and reshape C to shape [H * W, 2] and append ones to last dimension. The we get:
              C_ = [[-1.  -1.   1. ]
                    [-0.5 -1.   1. ]
                    [ 0.  -1.   1. ]
                    [ 0.5 -1.   1. ]
                    [ 1.  -1.   1. ]
                    [-1.  -0.5  1. ]
                    [-0.5 -0.5  1. ]
                    [ 0.  -0.5  1. ]
                    [ 0.5 -0.5  1. ]
                    [ 1.  -0.5  1. ]
                    [-1.   0.   1. ]
                    [-0.5  0.   1. ]
                    [ 0.   0.   1. ]
                    [ 0.5  0.   1. ]
                    [ 1.   0.   1. ]
                    [-1.   0.5  1. ]
                    [-0.5  0.5  1. ]
                    [ 0.   0.5  1. ]
                    [ 0.5  0.5  1. ]
                    [ 1.   0.5  1. ]
                    [-1.   1.   1. ]
                    [-0.5  1.   1. ]
                    [ 0.   1.   1. ]
                    [ 0.5  1.   1. ]
                    [ 1.   1.   1. ]]
          Step3:
              Compute output by equation $$Output[i] = C_ * Theta[i]^T$$

    Args:
        theta (Variable): A batch of affine transform parameters with shape [N, 2, 3].
        out_shape (Variable | list | tuple): The shape of target output with format [N, C, H, W].
        out_shape can be a Variable or a list or tuple.
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.

    Returns:
        Variable: The output with shape [N, H, W, 2].

    Raises:
        ValueError: If the type of arguments is not supported.

    Examples:

        .. code-block:: python
            theta = fluid.layers.data(name="x", shape=[2, 3], dtype="float32")
            out_shape = fluid.layers.data(name="y", shape=[-1], dtype="float32")
            data = fluid.layers.affine_grid(theta, out_shape)

            # or
            data = fluid.layers.affine_grid(theta, [5, 3, 28, 28])

    """
    helper = LayerHelper('affine_grid')

    if not (isinstance(out_shape, list) or isinstance(out_shape, tuple) or \
        isinstance(out_shape, Variable)):
        raise ValueError("The out_shape should be a list, tuple or Variable.")

    if not isinstance(theta, Variable):
        raise ValueError("The theta should be a Variable.")

    out = helper.create_variable_for_type_inference(theta.dtype)
    ipts = {'Theta': theta}
    attrs = {}
    if isinstance(out_shape, Variable):
        ipts['OutputShape'] = out_shape
    else:
        attrs['output_shape'] = out_shape

    helper.append_op(
        type='affine_grid',
        inputs=ipts,
        outputs={'Output': out},
        attrs=None if len(attrs) == 0 else attrs)
    return out


6550 6551 6552 6553 6554 6555 6556 6557
def rank_loss(label, left, right, name=None):
    """
    **Rank loss layer for RankNet**

    RankNet(http://icml.cc/2015/wp-content/uploads/2015/06/icml_ranking.pdf)
    is a pairwise ranking model with a training sample consisting of a pair
    of documents, A and B. Label P indicates whether A is ranked higher than B
    or not:
M
minqiyang 已提交
6558

6559 6560
    P = {0, 1} or {0, 0.5, 1}, where 0.5 means that there is no information
    about the rank of the input pair.
M
minqiyang 已提交
6561

6562 6563 6564 6565
    Rank loss layer takes three inputs: left (o_i), right (o_j) and
    label (P_{i,j}). The inputs respectively represent RankNet's output scores
    for documents A and B and the value of label P. The following equation
    computes rank loss C_{i,j} from the inputs:
M
minqiyang 已提交
6566

6567 6568 6569 6570 6571
    $$
      C_{i,j} = -\tilde{P_{ij}} * o_{i,j} + \log(1 + e^{o_{i,j}}) \\
      o_{i,j} =  o_i - o_j  \\
      \tilde{P_{i,j}} = \left \{0, 0.5, 1 \right \} \ or \ \left \{0, 1 \right \}
    $$
M
minqiyang 已提交
6572 6573 6574

    Rank loss layer takes batch inputs with size batch_size (batch_size >= 1).

6575 6576 6577 6578 6579 6580 6581 6582 6583 6584 6585 6586 6587 6588 6589 6590 6591 6592 6593 6594 6595 6596 6597 6598 6599 6600 6601 6602 6603 6604 6605 6606 6607 6608 6609
    Args:
        label (Variable): Indicats whether A ranked higher than B or not.
        left (Variable): RankNet's output score for doc A.
        right (Variable): RankNet's output score for doc B.
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.

    Returns:
        list: The value of rank loss.

    Raises:
        ValueError: Any of label, left, and right is not a variable.

    Examples:

        .. code-block:: python

            label = fluid.layers.data(name="label", shape=[4, 1], dtype="float32")
            left = fluid.layers.data(name="left", shape=[4, 1], dtype="float32")
            right = fluid.layers.data(name="right", shape=[4, 1], dtype="float32")
            out = fluid.layers.rank_loss(label, left, right)


    """
    helper = LayerHelper('rank_loss', **locals())

    if not (isinstance(label, Variable)):
        raise ValueError("The label should be a Variable")

    if not (isinstance(left, Variable)):
        raise ValueError("The left should be a Variable")

    if not (isinstance(right, Variable)):
        raise ValueError("The right should be a Variable")

X
Xin Pan 已提交
6610
    out = helper.create_variable_for_type_inference("float32")
6611 6612 6613 6614 6615 6616 6617 6618

    helper.append_op(
        type='rank_loss',
        inputs={"Label": label,
                "Left": left,
                "Right": right},
        outputs={'Out': out})
    return out
6619 6620


M
minqiyang 已提交
6621 6622
def margin_rank_loss(label, left, right, margin=0.1, name=None):
    """
M
minqiyang 已提交
6623
    Margin Ranking Loss Layer for ranking problem,
M
minqiyang 已提交
6624
    which compares left score and right score passed in.
M
minqiyang 已提交
6625
    The ranking loss can be defined as following equation:
M
minqiyang 已提交
6626 6627 6628 6629 6630 6631

    .. math::

        rank\_loss &= max(0, -label * (left - right) + margin)

    Args:
M
minqiyang 已提交
6632
       label (Variable): Indicates whether the left is ranked higher than the right or not.
M
minqiyang 已提交
6633 6634
       left (Variable): Ranking score for left.
       right (Variable): Ranking score for right.
M
minqiyang 已提交
6635
       margin (float): Indicates the given margin.
M
minqiyang 已提交
6636 6637 6638
       name (str|None): A name for this layer (optional). If set None, the layer
                       will be named automatically.
    Returns:
M
minqiyang 已提交
6639
       Variable: The ranking loss.
M
minqiyang 已提交
6640
    Raises:
M
minqiyang 已提交
6641
       ValueError: Any of label, left, and right is not a Variable.
M
minqiyang 已提交
6642 6643 6644 6645 6646 6647 6648
    Examples:
        .. code-block:: python
           label = fluid.layers.data(name="label", shape=[4, 1], dtype="float32")
           left = fluid.layers.data(name="left", shape=[4, 1], dtype="float32")
           right = fluid.layers.data(name="right", shape=[4, 1], dtype="float32")
           out = fluid.layers.margin_rank_loss(label, left, right)
    """
M
minqiyang 已提交
6649
    helper = LayerHelper('margin_rank_loss', **locals())
M
minqiyang 已提交
6650 6651 6652 6653 6654 6655
    if not isinstance(label, Variable):
        raise ValueError("The label should be a Variable.")
    if not isinstance(left, Variable):
        raise ValueError("The left should be a Variable.")
    if not isinstance(right, Variable):
        raise ValueError("The right should be a Variable.")
X
Xin Pan 已提交
6656 6657
    out = helper.create_variable_for_type_inference(left.dtype)
    act = helper.create_variable_for_type_inference(left.dtype)
M
minqiyang 已提交
6658 6659 6660 6661 6662 6663 6664 6665 6666 6667 6668
    helper.append_op(
        type='margin_rank_loss',
        inputs={"Label": label,
                "X1": left,
                "X2": right},
        outputs={'Out': out,
                 'Activated': act},
        attrs={'margin': margin})
    return out


W
whs 已提交
6669 6670 6671 6672 6673 6674 6675 6676 6677 6678 6679 6680 6681 6682
def pad2d(input,
          paddings=[0, 0, 0, 0],
          mode='constant',
          pad_value=0.0,
          data_format="NCHW",
          name=None):
    """
    Pad 2-d images accordding to 'paddings' and 'mode'.
    If mode is 'reflect', paddings[0] and paddings[1] must be no greater
    than height-1. And the width dimension has the same condition.

    Example:

      Given that X is a channel of image from input:
M
minqiyang 已提交
6683

W
whs 已提交
6684 6685
      X = [[1, 2, 3],
           [4, 5, 6]]
M
minqiyang 已提交
6686

W
whs 已提交
6687
      Case 0:
M
minqiyang 已提交
6688

W
whs 已提交
6689 6690 6691
        paddings = [0, 1, 2, 3],
        mode = 'constant'
        pad_value = 0
M
minqiyang 已提交
6692

W
whs 已提交
6693 6694 6695
        Out = [[0, 0, 1, 2, 3, 0, 0, 0]
               [0, 0, 4, 5, 6, 0, 0, 0]
               [0, 0, 0, 0, 0, 0, 0, 0]]
M
minqiyang 已提交
6696

W
whs 已提交
6697
      Case 1:
M
minqiyang 已提交
6698

W
whs 已提交
6699 6700
        paddings = [0, 1, 2, 1],
        mode = 'reflect'
M
minqiyang 已提交
6701

W
whs 已提交
6702 6703 6704
        Out = [[3, 2, 1, 2, 3, 2]
               [6, 5, 4, 5, 6, 5]
               [3, 2, 1, 2, 3, 2]]
M
minqiyang 已提交
6705

W
whs 已提交
6706
      Case 2:
M
minqiyang 已提交
6707

W
whs 已提交
6708 6709
        paddings = [0, 1, 2, 1],
        mode = 'edge'
M
minqiyang 已提交
6710

W
whs 已提交
6711 6712 6713
        Out = [[1, 1, 1, 2, 3, 3]
               [4, 4, 4, 5, 6, 6]
               [4, 4, 4, 5, 6, 6]]
M
minqiyang 已提交
6714 6715


W
whs 已提交
6716 6717 6718 6719 6720 6721 6722 6723 6724 6725 6726 6727 6728 6729 6730 6731 6732 6733 6734 6735 6736 6737 6738 6739 6740 6741
    Args:
        input (Variable): The input image with [N, C, H, W] format or [N, H, W, C] format.
        paddings (tuple|list): The padding size. If padding is a tuple, it must
            contain four integers, (padding_top, padding_bottom, padding_left, padding_right).
            Default: padding = [0, 0, 0, 0].
        mode (str): Three modes: constant(default), reflect, edge. Default: constant
        pad_value (float32): The value to fill the padded areas in constant mode. Default: 0
        data_format (str): An optional string from: "NHWC", "NCHW". Specify the data format of
                           the input data.
                           Default: "NCHW"
        name (str|None): A name for this layer(optional). If set None, the layer
            will be named automatically.

    Returns:
        Variable: The tensor variable padded accordding to paddings and mode.


    Examples:
        .. code-block:: python

          data = fluid.layers.data(name='data', shape=[3, 32, 32], dtype='float32')
          result = fluid.layers.pad2d(input=data, padding=[1,2,3,4], mode='reflect')
    """

    helper = LayerHelper('pad2d', **locals())
    dtype = helper.input_dtype(input_param_name='input')
X
Xin Pan 已提交
6742
    out = helper.create_variable_for_type_inference(dtype)
W
whs 已提交
6743 6744 6745 6746 6747 6748 6749 6750 6751 6752 6753 6754 6755 6756
    helper.append_op(
        type='pad2d',
        inputs={'X': input},
        outputs={"Out": out},
        attrs={
            'paddings': paddings,
            'mode': mode,
            'pad_value': pad_value,
            'data_frmat': data_format
        })

    return out


6757 6758 6759 6760 6761 6762 6763 6764 6765 6766 6767 6768 6769 6770
@templatedoc()
def elu(x, alpha=1.0, name=None):
    """
    ${comment}
    Args:
        x(${x_type}): ${x_comment}
        alpha(${alpha_type}|1.0): ${alpha_comment}
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.

    Returns:
        output(${out_type}): ${out_comment}
    """
    helper = LayerHelper('elu', **locals())
X
Xin Pan 已提交
6771
    out = helper.create_variable_for_type_inference(dtype=x.dtype)
6772 6773 6774 6775 6776 6777 6778 6779 6780 6781 6782 6783 6784 6785 6786 6787 6788 6789 6790 6791 6792 6793
    helper.append_op(
        type='elu',
        inputs={'X': x},
        outputs={'Out': out},
        attrs={'alpha': alpha})
    return out


@templatedoc()
def relu6(x, threshold=6.0, name=None):
    """
    ${comment}
    Args:
        x(${x_type}): ${x_comment}
        threshold(${threshold_type}|6.0): ${threshold_comment}
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.

    Returns:
        output(${out_type}): ${out_comment}
    """
    helper = LayerHelper('relu6', **locals())
X
Xin Pan 已提交
6794
    out = helper.create_variable_for_type_inference(dtype=x.dtype)
6795 6796 6797 6798 6799 6800 6801 6802 6803 6804 6805 6806 6807 6808 6809 6810 6811 6812 6813 6814 6815 6816
    helper.append_op(
        type='relu6',
        inputs={'X': x},
        outputs={'Out': out},
        attrs={'threshold': threshold})
    return out


@templatedoc()
def pow(x, factor=1.0, name=None):
    """
    ${comment}
    Args:
        x(${x_type}): ${x_comment}
        factor(${factor_type}|1.0): ${factor_comment}
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.

    Returns:
        output(${out_type}): ${out_comment}
    """
    helper = LayerHelper('pow', **locals())
X
Xin Pan 已提交
6817
    out = helper.create_variable_for_type_inference(dtype=x.dtype)
6818 6819 6820 6821 6822 6823 6824 6825 6826 6827 6828 6829 6830 6831 6832 6833 6834 6835 6836 6837 6838 6839 6840
    helper.append_op(
        type='pow',
        inputs={'X': x},
        outputs={'Out': out},
        attrs={'factor': factor})
    return out


@templatedoc()
def stanh(x, scale_a=2.0 / 3.0, scale_b=1.7159, name=None):
    """
    ${comment}
    Args:
        x(${x_type}): ${x_comment}
        scale_a(${scale_a_type}|2.0 / 3.0): ${scale_a_comment}
        scale_b(${scale_b_type}|1.7159): ${scale_b_comment}
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.

    Returns:
        output(${out_type}): ${out_comment}
    """
    helper = LayerHelper('stanh', **locals())
X
Xin Pan 已提交
6841
    out = helper.create_variable_for_type_inference(dtype=x.dtype)
6842 6843 6844 6845 6846 6847 6848 6849 6850 6851 6852 6853 6854 6855 6856 6857 6858 6859 6860 6861 6862 6863 6864 6865
    helper.append_op(
        type='stanh',
        inputs={'X': x},
        outputs={'Out': out},
        attrs={'scale_a': scale_a,
               'scale_b': scale_b})
    return out


@templatedoc()
def hard_sigmoid(x, slope=0.2, offset=0.5, name=None):
    """
    ${comment}
    Args:
        x(${x_type}): ${x_comment}
        slope(${slope_type}|0.2): ${slope_comment}
        offset(${offset_type}|0.5): ${offset_comment}
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.

    Returns:
        output(${out_type}): ${out_comment}
    """
    helper = LayerHelper('hard_sigmoid', **locals())
X
Xin Pan 已提交
6866
    out = helper.create_variable_for_type_inference(dtype=x.dtype)
6867 6868 6869 6870 6871 6872 6873 6874 6875 6876 6877 6878 6879 6880 6881 6882 6883 6884 6885 6886 6887 6888 6889
    helper.append_op(
        type='hard_sigmoid',
        inputs={'X': x},
        outputs={'Out': out},
        attrs={'slope': slope,
               'offset': offset})
    return out


@templatedoc()
def swish(x, beta=1.0, name=None):
    """
    ${comment}
    Args:
        x(${x_type}): ${x_comment}
        beta(${beta_type}|1.0): ${beta_comment}
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.

    Returns:
        output(${out_type}): ${out_comment}
    """
    helper = LayerHelper('swish', **locals())
X
Xin Pan 已提交
6890
    out = helper.create_variable_for_type_inference(dtype=x.dtype)
6891 6892 6893 6894 6895 6896 6897 6898
    helper.append_op(
        type='swish',
        inputs={'X': x},
        outputs={'Out': out},
        attrs={'slope': beta})
    return out


J
jerrywgz 已提交
6899 6900 6901 6902 6903 6904 6905 6906 6907 6908 6909 6910 6911 6912
def prelu(x, mode, param_attr=None, name=None):
    """
    Equation:

        y = \max(0, x) + alpha \min(0, x)

    Args:
        x (Variable): The input tensor.
	  param_attr(ParamAttr|None): The parameter attribute for the learnable
                                    weight (alpha).
        mode (string): The mode for weight sharing
		       all: all elements share same weight
 		       channel:elements in a channel share same weight
 		       element:each element has a weight
W
whs 已提交
6913
	name(str|None): A name for this layer(optional). If set None, the layer
M
minqiyang 已提交
6914
                        will be named automatically.
J
jerrywgz 已提交
6915 6916 6917 6918 6919 6920 6921 6922 6923 6924 6925 6926 6927 6928 6929 6930 6931 6932 6933 6934 6935 6936

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

    Examples:

        .. code-block:: python

         x = fluid.layers.data(name="x", shape=[10,10], dtype="float32")
            mode = 'channel'
            output = fluid.layers.prelu(x,mode)
    """
    helper = LayerHelper('prelu', **locals())
    if mode not in ['all', 'channel', 'element']:
        raise ValueError('mode should be one of all, channel, element.')
    alpha_shape = [1]
    if mode == 'channel':
        alpha_shape = [1, x.shape[1], 1, 1]
    elif mode == 'element':
        alpha_shape = x.shape
    dtype = helper.input_dtype(input_param_name='x')
    alpha = helper.create_parameter(
Q
Qiao Longfei 已提交
6937
        attr=helper.param_attr,
J
jerrywgz 已提交
6938 6939 6940 6941
        shape=alpha_shape,
        dtype='float32',
        is_bias=False,
        default_initializer=Constant(1.0))
X
Xin Pan 已提交
6942
    out = helper.create_variable_for_type_inference(dtype)
J
jerrywgz 已提交
6943 6944 6945 6946 6947 6948 6949 6950 6951
    helper.append_op(
        type="prelu",
        inputs={"X": x,
                'Alpha': alpha},
        attrs={"mode": mode},
        outputs={"Out": out})
    return out


6952 6953 6954 6955 6956 6957 6958 6959 6960 6961
@templatedoc()
def brelu(x, t_min=0.0, t_max=24.0, name=None):
    """
    ${comment}
    Args:
        x(${x_type}): ${x_comment}
        t_min(${t_min_type}|0.0): ${t_min_comment}
        t_max(${t_max_type}|24.0): ${t_max_comment}
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.
6962
    Returns:
6963
        output(${out_type}): ${out_comment}
6964 6965 6966 6967 6968 6969 6970

    Examples:

        .. code-block:: python

        x = fluid.layers.data(name="x", shape=[2,3,16,16], dtype="float32")
        y = fluid.layers.brelu(x, t_min=1.0, t_max=20.0)
6971 6972
    """
    helper = LayerHelper('brelu', **locals())
X
Xin Pan 已提交
6973
    out = helper.create_variable_for_type_inference(dtype=x.dtype)
6974 6975 6976 6977 6978 6979 6980 6981 6982 6983 6984 6985 6986 6987 6988 6989 6990 6991
    helper.append_op(
        type='brelu',
        inputs={'X': x},
        outputs={'Out': out},
        attrs={'t_min': t_min,
               't_max': t_max})
    return out


@templatedoc()
def leaky_relu(x, alpha=0.02, name=None):
    """
    ${comment}
    Args:
        x(${x_type}): ${x_comment}
        alpha(${alpha_type}|0.02): ${alpha_comment}
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.
6992
    Returns:
6993
        output(${out_type}): ${out_comment}
6994 6995 6996 6997 6998 6999 7000

    Examples:

        .. code-block:: python

        x = fluid.layers.data(name="x", shape=[2,3,16,16], dtype="float32")
        y = fluid.layers.leaky_relu(x, alpha=0.01)
7001 7002
    """
    helper = LayerHelper('leaky_relu', **locals())
X
Xin Pan 已提交
7003
    out = helper.create_variable_for_type_inference(dtype=x.dtype)
7004 7005 7006 7007 7008 7009 7010 7011 7012 7013 7014 7015 7016 7017 7018 7019 7020
    helper.append_op(
        type='leaky_relu',
        inputs={'X': x},
        outputs={'Out': out},
        attrs={'alpha': alpha})
    return out


@templatedoc()
def soft_relu(x, threshold=40.0, name=None):
    """
    ${comment}
    Args:
        x(${x_type}): ${x_comment}
        threshold(${threshold_type}|40.0): ${threshold_comment}
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.
7021
    Returns:
7022
        output(${out_type}): ${out_comment}
7023 7024 7025 7026 7027 7028 7029

    Examples:

        .. code-block:: python

        x = fluid.layers.data(name="x", shape=[2,3,16,16], dtype="float32")
        y = fluid.layers.soft_relu(x, threshold=20.0)
7030 7031
    """
    helper = LayerHelper('soft_relu', **locals())
X
Xin Pan 已提交
7032
    out = helper.create_variable_for_type_inference(dtype=x.dtype)
7033 7034 7035 7036 7037 7038 7039 7040
    helper.append_op(
        type='soft_relu',
        inputs={'X': x},
        outputs={'Out': out},
        attrs={'threshold': threshold})
    return out


7041 7042 7043 7044 7045 7046 7047 7048 7049 7050 7051 7052 7053
def flatten(x, axis=1, name=None):
    """
    **Flatten layer**
    Flattens the input tensor into a 2D matrix.

    Examples:
    Case 1:
      Given
        X.shape = (3, 100, 100, 4)
      and
        axis = 2
      We get:
        Out.shape = (3 * 100, 4 * 100)
7054

7055 7056 7057 7058 7059 7060 7061 7062 7063 7064
    Case 2:
      Given
        X.shape = (3, 100, 100, 4)
      and
        axis = 0
      We get:
        Out.shape = (1, 3 * 100 * 100 * 4)

    Args:
        x (Variable): A tensor of rank >= axis.
7065 7066
        axis (int): Indicate up to which input dimensions (exclusive) should
                    be flattened to the outer dimension of the output.
7067 7068 7069 7070 7071 7072 7073 7074 7075 7076 7077 7078 7079 7080 7081
                    The value for axis must be in the range [0, R], where R
                    is the rank of the input tensor. When axis = 0, the shape
                    of the output tensor is (1, (d_0 X d_1 ... d_n), where the
                    shape of the input tensor is (d_0, d_1, ... d_n).
        name(str|None): A name for this layer(optional). If set None, the layer
                        will be named automatically.

    Returns:
        Variable: A 2D tensor with the contents of the input tensor, with input
                  dimensions up to axis flattened to the outer dimension of
                  the output and remaining input dimensions flattened into the
                  inner dimension of the output.

    Raises:
        ValueError: If x is not a variable.
7082
        ValueError: If axis is not in range [0, rank(x)].
7083 7084 7085 7086 7087 7088 7089 7090 7091 7092 7093 7094 7095 7096 7097 7098

    Examples:

        .. code-block:: python

            x = fluid.layers.data(name="x", shape=[4, 4, 3], dtype="float32")
            out = fluid.layers.flatten(x=x, axis=2)
    """
    helper = LayerHelper('flatten', **locals())

    if not (isinstance(x, Variable)):
        raise ValueError("The input x should be a Variable")

    if not (isinstance(axis, int)) or axis > len(x.shape) or axis < 0:
        raise ValueError("The axis should be a int, and in range [0, rank(x)]")

X
Xin Pan 已提交
7099 7100
    out = helper.create_variable_for_type_inference(x.dtype)
    x_shape = helper.create_variable_for_type_inference(x.dtype)
7101
    helper.append_op(
7102
        type='flatten2',
7103
        inputs={"X": x},
7104 7105
        outputs={'Out': out,
                 'XShape': x_shape},
7106 7107
        attrs={"axis": axis})
    return out
X
Xin Pan 已提交
7108 7109


C
chenweihang 已提交
7110
def sequence_enumerate(input, win_size, pad_value=0, name=None):
C
chenweihang 已提交
7111
    """
C
chenweihang 已提交
7112
    Generate a new sequence for the input index sequence, which enumerates all the
M
minqiyang 已提交
7113
    sub-sequences with length `win_size` of the input.
C
chenweihang 已提交
7114 7115
    The enumerated sequence has the same 1st dimension with variable `input`, and
    the 2nd dimension is `win_size`, padded by `pad_value` if necessary in generation.
M
minqiyang 已提交
7116

C
chenweihang 已提交
7117 7118 7119 7120
    Examples:
    Case 1:
      Input:
        X.lod = [[0, 3, 5]]
7121
        X.data = [[1], [2], [3], [4], [5]]
C
chenweihang 已提交
7122 7123 7124 7125 7126 7127
        X.dims = [5, 1]
      Attrs:
        win_size = 2
        pad_value = 0
      Output:
        Out.lod = [[0, 3, 5]]
7128
        Out.data = [[1, 2], [2, 3], [3, 0], [4, 5], [5, 0]]
C
chenweihang 已提交
7129 7130 7131
        Out.dims = [5, 2]

    Args:
C
chenweihang 已提交
7132 7133 7134
        input (Variable): The input variable which is a index sequence.
        win_size (int): The window size for enumerating all sub-sequences.
        pad_value (int): The padding value, default 0.
C
chenweihang 已提交
7135 7136 7137 7138 7139 7140 7141 7142 7143 7144 7145

    Returns:
        Variable: The enumerate sequence variable which is a LoDTensor.

    Examples:
        .. code-block:: python

            x = fluid.layers.data(shape[30, 1], dtype='int32', lod_level=1)
            out = fluid.layers.sequence_enumerate(input=x, win_size=3, pad_value=0)
    """
    helper = LayerHelper('sequence_enumerate', **locals())
X
Xin Pan 已提交
7146 7147
    out = helper.create_variable_for_type_inference(
        helper.input_dtype(), stop_gradient=True)
C
chenweihang 已提交
7148 7149 7150 7151 7152 7153
    helper.append_op(
        type='sequence_enumerate',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={'win_size': win_size,
               'pad_value': pad_value})
M
minqiyang 已提交
7154
    return out
7155

7156

S
sneaxiy 已提交
7157 7158 7159 7160 7161 7162 7163 7164 7165
def sequence_mask(x, maxlen=None, dtype='int64', name=None):
    """
    **SequenceMask Layer**

    This layer outputs a mask according to the input :code:`x` and
    :code:`maxlen` with data type of :code:`dtype`.

    Supposing :code:`x` is a Tensor with shape [d_1, d_2, ..., d_n], the
    :code:`y` is a mask with shape [d_1, d_2, ..., d_n, maxlen], where:
7166

S
sneaxiy 已提交
7167
    .. math::
7168

S
sneaxiy 已提交
7169 7170 7171
        y(i_1, i_2,..., i_n, j) = (j < x(i_1, i_2,..., i_n))

    Args:
7172
        x (Variable): Input tensor of sequence_mask layer,
S
sneaxiy 已提交
7173 7174 7175 7176
                      whose elements are integers less than :code:`maxlen`.
        maxlen (int|None): Maximum length of the sequence. If :code:`maxlen`
                           is None, it would be replace with :math:`max(x)`.
        dtype (np.dtype|core.VarDesc.VarType|str): Data type of the output.
7177 7178 7179
        name (str|None): A name for this layer(optional). If set None, the
                         layer will be named automatically.

S
sneaxiy 已提交
7180 7181
    Returns:
        Variable: The output sequence mask.
7182

S
sneaxiy 已提交
7183 7184
    """

Q
qingqing01 已提交
7185
    helper = LayerHelper('sequence_mask', **locals())
S
sneaxiy 已提交
7186
    if name is None:
X
Xin Pan 已提交
7187
        out = helper.create_variable_for_type_inference(dtype=dtype)
S
sneaxiy 已提交
7188
    else:
X
Xin Pan 已提交
7189
        out = helper.create_variable_for_type_inference(dtype=dtype, name=name)
S
sneaxiy 已提交
7190

Q
qingqing01 已提交
7191 7192 7193
    helper.append_op(
        type='sequence_mask',
        inputs={'X': [x]},
S
sneaxiy 已提交
7194 7195
        outputs={'Y': out},
        attrs={
7196
            'maxlen': maxlen if maxlen is not None else -1,
S
sneaxiy 已提交
7197 7198 7199
            'out_dtype': out.dtype
        })
    return out
S
sneaxiy 已提交
7200 7201


X
Xin Pan 已提交
7202
def stack(x, axis=0):
S
sneaxiy 已提交
7203 7204 7205 7206
    """
    **Stack Layer**

    This layer stacks all of the input :code:`x` along axis.
7207 7208 7209 7210 7211 7212 7213

    Input :code:`x` can be a single variable, a :code:`list` of variables,
    or a :code:`tuple` of variables. If :code:`x` is a :code:`list` or
    :code:`tuple`, the shapes of all these variables must be the same.
    Supposing the shape of each input is :math:`[d_0, d_1, ..., d_{n-1}]`,
    the shape of the output variable would be
    :math:`[d_0, d_1, ..., d_{axis}=len(x), ..., d_{n-1}]`.
S
sneaxiy 已提交
7214
    If :code:`axis` < 0, it would be replaced with :code:`axis+rank(x[0])+1`.
7215
    If :code:`axis` is None, it would be replaced with 0.
S
sneaxiy 已提交
7216 7217

    Args:
7218
        x (Variable|list(Variable)|tuple(Variable)): Input variables.
S
sneaxiy 已提交
7219
        axis (int|None): The axis along which all inputs are stacked.
7220

S
sneaxiy 已提交
7221 7222
    Returns:
        Variable: The stacked variable.
7223

S
sneaxiy 已提交
7224 7225
    """

X
Xin Pan 已提交
7226 7227 7228 7229 7230 7231
    helper = LayerHelper('stack', **locals())
    axis = 0 if axis is None else axis

    if not isinstance(x, list) and not isinstance(x, tuple):
        x = [x]

X
Xin Pan 已提交
7232
    out = helper.create_variable_for_type_inference(x[0].dtype)
X
Xin Pan 已提交
7233
    helper.append_op(
S
sneaxiy 已提交
7234 7235
        type='stack', inputs={'X': x}, outputs={'Y': out},
        attrs={'axis': axis})
7236

X
Xin Pan 已提交
7237
    return out
D
dzhwinter 已提交
7238 7239 7240 7241 7242 7243 7244


def unstack(x, axis=0, num=None):
    """
    **UnStack Layer**

    This layer unstacks input :code:`x` into several tensors along axis.
M
minqiyang 已提交
7245

D
dzhwinter 已提交
7246 7247 7248
    If :code:`axis` < 0, it would be replaced with :code:`axis+rank(x)`.
    If :code:`num` is None, it would be inferred from :code:`x.shape[axis]`,
    and if :code:`x.shape[axis]` <= 0 or is unknown, :code:`ValueError` is
M
minqiyang 已提交
7249
    raised.
D
dzhwinter 已提交
7250 7251

    Args:
M
minqiyang 已提交
7252
        x (Variable): Input variable.
D
dzhwinter 已提交
7253 7254
        axis (int): The axis along which the input is unstacked.
        num (int|None): The number of output variables.
M
minqiyang 已提交
7255

D
dzhwinter 已提交
7256 7257
    Returns:
        list(Variable): The unstacked variables.
M
minqiyang 已提交
7258

D
dzhwinter 已提交
7259 7260 7261 7262 7263 7264 7265 7266 7267 7268 7269
    """

    helper = LayerHelper('unstack', **locals())
    if num is None:
        if axis is None or x.shape[axis] <= 0:
            raise ValueError('unknown unstack number')
        else:
            num = x.shape[axis]

    outs = []
    for _ in num:
X
Xin Pan 已提交
7270
        outs.append(helper.create_variable_for_type_inference(x.dtype))
D
dzhwinter 已提交
7271 7272 7273 7274 7275 7276 7277 7278

    helper.append_op(
        type='unstack',
        inputs={'X': [x]},
        outputs={'Y': outs},
        attrs={'axis': axis,
               'num': num})
    return outs
W
whs 已提交
7279 7280 7281 7282 7283 7284 7285 7286 7287 7288 7289 7290


def expand(x, expand_times, name=None):
    """Expand operator tiles the input by given times number. You should set times
    number for each dimension by providing attribute 'expand_times'. The rank of X
    should be in [1, 6]. Please note that size of 'expand_times' must be the same
    with X's rank. Following is a using case:


    .. code-block:: text

        Input(X) is a 3-D tensor with shape [2, 3, 1]:
M
minqiyang 已提交
7291

W
whs 已提交
7292 7293 7294 7295
                [
                   [[1], [2], [3]],
                   [[4], [5], [6]]
                ]
M
minqiyang 已提交
7296

W
whs 已提交
7297
        Attr(expand_times):  [1, 2, 2]
M
minqiyang 已提交
7298

W
whs 已提交
7299
        Output(Out) is a 3-D tensor with shape [2, 6, 2]:
M
minqiyang 已提交
7300

W
whs 已提交
7301 7302 7303 7304
                [
                    [[1, 1], [2, 2], [3, 3], [1, 1], [2, 2], [3, 3]],
                    [[4, 4], [5, 5], [6, 6], [4, 4], [5, 5], [6, 6]]
                ]
M
minqiyang 已提交
7305

W
whs 已提交
7306 7307 7308 7309 7310 7311 7312 7313 7314 7315 7316 7317 7318 7319 7320 7321
    Args:
        x (Variable): A tensor with rank in [1, 6].
        expand_times (list|tuple): Expand times number for each dimension.

    Returns:
        Variable: The expanded variable which is a LoDTensor. After expanding, size of each dimension of Output(Out) is equal to ithe size of the corresponding dimension of Input(X) multiplying the corresponding value given by expand_times.


    Examples:
        .. code-block:: python

            x = fluid.layers.data(name='x', shape=[10], dtype='float32')
            out = fluid.layers.expand(x=x, expand_times=[1, 2, 2])
    """
    helper = LayerHelper('expand', input=x, **locals())
    dtype = helper.input_dtype(input_param_name='x')
X
Xin Pan 已提交
7322
    out = helper.create_variable_for_type_inference(dtype)
W
whs 已提交
7323 7324 7325 7326 7327 7328
    helper.append_op(
        type='expand',
        inputs={'X': x},
        outputs={'Out': out},
        attrs={'expand_times': expand_times})
    return out
S
sneaxiy 已提交
7329 7330


G
fix  
gongweibao 已提交
7331 7332 7333
from paddle.fluid.framework import convert_np_dtype_to_dtype_


G
gongweibao 已提交
7334
@templatedoc()
G
fix  
gongweibao 已提交
7335 7336 7337 7338 7339 7340 7341 7342 7343
def uniform_random_batch_size_like(input,
                                   shape,
                                   dtype='float32',
                                   input_dim_idx=0,
                                   output_dim_idx=0,
                                   min=-1.0,
                                   max=1.0,
                                   seed=0):
    """
G
gongweibao 已提交
7344
    ${comment}
G
fix  
gongweibao 已提交
7345 7346

    Args:
G
gongweibao 已提交
7347 7348 7349
        input (Variable): ${input_comment}
        shape (tuple|list): ${shape_comment}
        input_dim_idx (Int): ${input_dim_idx_comment}
G
gongweibao 已提交
7350
        output_dim_idx (Int): ${output_dim_idx_comment}
G
gongweibao 已提交
7351 7352 7353
        min (Float): ${min_comment}
        max (Float): ${max_comment}
        seed (Int): ${seed_comment}
G
fix  
gongweibao 已提交
7354 7355
        dtype(np.dtype|core.VarDesc.VarType|str): The type of data : float32, float_16, int etc
    Returns:
G
gongweibao 已提交
7356
        out (Variable): ${out_comment}
G
fix  
gongweibao 已提交
7357 7358 7359 7360

    """

    helper = LayerHelper('uniform_random_batch_size_like', **locals())
X
Xin Pan 已提交
7361
    out = helper.create_variable_for_type_inference(dtype)
G
fix  
gongweibao 已提交
7362 7363 7364 7365 7366 7367 7368 7369 7370 7371 7372 7373 7374 7375 7376 7377
    c_dtype = convert_np_dtype_to_dtype_(dtype)
    helper.append_op(
        type='uniform_random_batch_size_like',
        inputs={'Input': input},
        outputs={'Out': out},
        attrs={
            'shape': shape,
            'input_dim_idx': input_dim_idx,
            'output_dim_idx': output_dim_idx,
            'min': min,
            'max': max,
            'seed': seed,
            'dtype': c_dtype
        })

    return out
G
fix  
gongweibao 已提交
7378 7379


G
gongweibao 已提交
7380
@templatedoc()
X
Xin Pan 已提交
7381
def gaussian_random(shape, mean=0.0, std=1.0, seed=0, dtype='float32'):
G
fix  
gongweibao 已提交
7382
    """
G
gongweibao 已提交
7383
    ${comment}
G
fix  
gongweibao 已提交
7384 7385

    Args:
G
gongweibao 已提交
7386 7387 7388 7389
        shape (tuple|list): ${shape_comment}
        mean (Float): ${mean_comment}
        std (Float): ${std_comment}
        seed (Int): ${seed_comment}
G
fix  
gongweibao 已提交
7390 7391 7392
        dtype(np.dtype|core.VarDesc.VarType|str): Output data type.

    Returns:
G
gongweibao 已提交
7393
        out (Variable): ${out_comment}
G
fix  
gongweibao 已提交
7394 7395 7396 7397

    """

    helper = LayerHelper('gaussian_random', **locals())
X
Xin Pan 已提交
7398
    out = helper.create_variable_for_type_inference(dtype)
G
fix  
gongweibao 已提交
7399 7400 7401 7402 7403 7404 7405 7406 7407 7408
    c_dtype = convert_np_dtype_to_dtype_(dtype)
    helper.append_op(
        type='gaussian_random',
        outputs={'Out': out},
        attrs={
            'shape': shape,
            'mean': mean,
            'std': std,
            'seed': seed,
            'dtype': c_dtype,
X
Xin Pan 已提交
7409
            'use_mkldnn': False
G
fix  
gongweibao 已提交
7410 7411 7412 7413 7414
        })

    return out


G
gongweibao 已提交
7415
@templatedoc()
G
fix  
gongweibao 已提交
7416
def sampling_id(x, min=0.0, max=1.0, seed=0, dtype='float32'):
G
fix  
gongweibao 已提交
7417
    """
G
gongweibao 已提交
7418
    ${comment}
G
fix  
gongweibao 已提交
7419 7420

    Args:
G
gongweibao 已提交
7421 7422 7423 7424
        x (Variable): ${x_comment}
        min (Float): ${min_comment}
        max (Float): ${max_comment}
        seed (Float): ${seed_comment}
G
fix  
gongweibao 已提交
7425
        dtype(np.dtype|core.VarDesc.VarType|str): The type of output data : float32, float_16, int etc
G
fix  
gongweibao 已提交
7426 7427

    Returns:
G
gongweibao 已提交
7428
        out (Variable): ${out_comment}
G
fix  
gongweibao 已提交
7429 7430 7431 7432

    """

    helper = LayerHelper('sampling_id', **locals())
X
Xin Pan 已提交
7433
    out = helper.create_variable_for_type_inference(dtype)
G
fix  
gongweibao 已提交
7434 7435 7436 7437 7438 7439 7440 7441 7442 7443 7444
    helper.append_op(
        type='sampling_id',
        inputs={'X': x},
        outputs={'Out': out},
        attrs={'min': min,
               'max': max,
               'seed': seed})

    return out


G
gongweibao 已提交
7445
@templatedoc()
G
fix  
gongweibao 已提交
7446 7447 7448 7449 7450 7451 7452 7453 7454
def gaussian_random_batch_size_like(input,
                                    shape,
                                    input_dim_idx=0,
                                    output_dim_idx=0,
                                    mean=0.0,
                                    std=1.0,
                                    seed=0,
                                    dtype='float32'):
    """
G
gongweibao 已提交
7455
    ${comment}
G
fix  
gongweibao 已提交
7456 7457

    Args:
G
gongweibao 已提交
7458 7459
        input (Variable): ${input_comment}
        shape (tuple|list): ${shape_comment}
G
gongweibao 已提交
7460
        input_dim_idx (Int): ${input_dim_idx_comment}
G
gongweibao 已提交
7461 7462 7463 7464
        output_dim_idx (Int): ${output_dim_idx_comment}
        mean (Float): ${mean_comment}
        std (Float): ${std_comment}
        seed (Int): ${seed_comment}
G
fix  
gongweibao 已提交
7465
        dtype(np.dtype|core.VarDesc.VarType|str): The type of output data : float32, float_16, int etc
G
fix  
gongweibao 已提交
7466 7467

    Returns:
G
gongweibao 已提交
7468
        out (Variable): ${out_comment}
G
fix  
gongweibao 已提交
7469 7470 7471
    """

    helper = LayerHelper('gaussian_random_batch_size_like', **locals())
X
Xin Pan 已提交
7472
    out = helper.create_variable_for_type_inference(dtype)
G
fix  
gongweibao 已提交
7473 7474 7475 7476 7477 7478 7479 7480 7481 7482 7483 7484 7485 7486 7487 7488 7489 7490
    c_dtype = convert_np_dtype_to_dtype_(dtype)
    helper.append_op(
        type='gaussian_random_batch_size_like',
        inputs={'Input': input},
        outputs={'Out': out},
        attrs={
            'shape': shape,
            'input_dim_idx': input_dim_idx,
            'output_dim_idx': output_dim_idx,
            'mean': mean,
            'std': std,
            'seed': seed,
            'dtype': c_dtype
        })

    return out


G
gongweibao 已提交
7491
@templatedoc()
X
Xin Pan 已提交
7492
def sum(x):
G
fix  
gongweibao 已提交
7493
    """
G
gongweibao 已提交
7494
    ${comment}
G
fix  
gongweibao 已提交
7495 7496

    Args:
G
gongweibao 已提交
7497
        x (Variable): ${x_comment}
G
fix  
gongweibao 已提交
7498 7499

    Returns:
G
gongweibao 已提交
7500
        out (Variable): ${out_comment}
G
fix  
gongweibao 已提交
7501 7502 7503
    """

    helper = LayerHelper('sum', **locals())
X
Xin Pan 已提交
7504 7505
    out = helper.create_variable_for_type_inference(
        dtype=helper.input_dtype('x'))
G
fix  
gongweibao 已提交
7506 7507 7508 7509
    helper.append_op(
        type='sum',
        inputs={'X': x},
        outputs={'Out': out},
X
Xin Pan 已提交
7510
        attrs={'use_mkldnn': False})
G
fix  
gongweibao 已提交
7511 7512 7513 7514

    return out


G
gongweibao 已提交
7515
@templatedoc()
G
fix  
gongweibao 已提交
7516 7517
def slice(input, axes, starts, ends):
    """
G
gongweibao 已提交
7518
    ${comment}
G
fix  
gongweibao 已提交
7519 7520

    Args:
G
gongweibao 已提交
7521 7522 7523 7524
        input (Variable): ${input_comment}.
        axes (List): ${axes_comment}
        starts (List): ${starts_comment}
        ends (List): ${ends_comment}
G
fix  
gongweibao 已提交
7525 7526

    Returns:
G
gongweibao 已提交
7527
        out (Variable): ${out_comment}
G
fix  
gongweibao 已提交
7528 7529 7530 7531

    """

    helper = LayerHelper('slice', **locals())
X
Xin Pan 已提交
7532 7533
    out = helper.create_variable_for_type_inference(
        dtype=helper.input_dtype('input'))
G
fix  
gongweibao 已提交
7534 7535 7536 7537 7538 7539 7540 7541 7542 7543 7544
    helper.append_op(
        type='slice',
        inputs={'Input': input},
        outputs={'Out': out},
        attrs={'axes': axes,
               'starts': starts,
               'ends': ends})

    return out


G
gongweibao 已提交
7545
@templatedoc()
G
fix  
gongweibao 已提交
7546 7547
def shape(input):
    """
G
gongweibao 已提交
7548
    ${comment}
G
fix  
gongweibao 已提交
7549 7550

    Args:
G
gongweibao 已提交
7551
        input (Variable): ${input_comment}
G
fix  
gongweibao 已提交
7552 7553

    Returns:
G
gongweibao 已提交
7554
        out (Variable): ${out_comment}
G
fix  
gongweibao 已提交
7555 7556 7557 7558

    """

    helper = LayerHelper('shape', **locals())
X
Xin Pan 已提交
7559 7560
    out = helper.create_variable_for_type_inference(
        dtype=helper.input_dtype('input'))
G
fix  
gongweibao 已提交
7561
    helper.append_op(
G
fix  
gongweibao 已提交
7562
        type='shape', inputs={'Input': input}, outputs={'Out': out})
G
fix  
gongweibao 已提交
7563 7564

    return out
G
merge  
gongweibao 已提交
7565 7566


S
sneaxiy 已提交
7567 7568 7569 7570 7571 7572 7573 7574
def _elementwise_op(helper):
    op_type = helper.layer_type
    x = helper.kwargs.get('x', None)
    y = helper.kwargs.get('y', None)
    assert x is not None, 'x cannot be None in {}'.format(op_type)
    assert y is not None, 'y cannot be None in {}'.format(op_type)
    axis = helper.kwargs.get('axis', -1)
    use_mkldnn = helper.kwargs.get('use_mkldnn', False)
S
sneaxiy 已提交
7575 7576
    name = helper.kwargs.get('name', None)
    if name is None:
X
Xin Pan 已提交
7577
        out = helper.create_variable_for_type_inference(dtype=x.dtype)
S
sneaxiy 已提交
7578 7579 7580
    else:
        out = helper.create_variable(
            name=name, dtype=x.dtype, persistable=False)
S
sneaxiy 已提交
7581

S
sneaxiy 已提交
7582 7583 7584 7585 7586 7587 7588 7589 7590 7591 7592
    helper.append_op(
        type=op_type,
        inputs={'X': x,
                'Y': y},
        outputs={'Out': out},
        attrs={'axis': axis,
               'use_mkldnn': use_mkldnn})
    return helper.append_activation(out)


@templatedoc()
S
sneaxiy 已提交
7593
def scale(x, scale=1.0, bias=0.0, bias_after_scale=True, act=None, name=None):
S
sneaxiy 已提交
7594 7595 7596 7597 7598 7599 7600 7601
    """
    ${comment}

    Args:
        x(${x_type}): ${x_comment}
        scale(${scale_type}): ${scale_comment}
        bias(${bias_type}): ${bias_comment}
        bias_after_scale(${bias_after_scale_type}): ${bias_after_scale_comment}
S
sneaxiy 已提交
7602
        act(basestring|None): Activation applied to the output.
M
minqiyang 已提交
7603
        name(basestring|None): Name of the output.
S
sneaxiy 已提交
7604 7605 7606 7607 7608 7609

    Returns:
        out(${out_type}): ${out_comment}
    """

    helper = LayerHelper('scale', **locals())
S
sneaxiy 已提交
7610
    if name is None:
X
Xin Pan 已提交
7611
        out = helper.create_variable_for_type_inference(dtype=x.dtype)
S
sneaxiy 已提交
7612 7613 7614
    else:
        out = helper.create_variable(
            name=name, dtype=x.dtype, persistable=False)
S
sneaxiy 已提交
7615 7616 7617 7618 7619 7620 7621 7622 7623 7624

    helper.append_op(
        type='scale',
        inputs={'X': x},
        outputs={'Out': out},
        attrs={
            'scale': float(scale),
            'bias': float(bias),
            'bias_after_scale': bias_after_scale
        })
S
sneaxiy 已提交
7625
    return helper.append_activation(out)
S
sneaxiy 已提交
7626 7627


X
Xin Pan 已提交
7628
def elementwise_add(x, y, axis=-1, act=None, name=None):
S
sneaxiy 已提交
7629 7630 7631
    return _elementwise_op(LayerHelper('elementwise_add', **locals()))


X
Xin Pan 已提交
7632
def elementwise_div(x, y, axis=-1, act=None, name=None):
S
sneaxiy 已提交
7633 7634 7635
    return _elementwise_op(LayerHelper('elementwise_div', **locals()))


X
Xin Pan 已提交
7636
def elementwise_sub(x, y, axis=-1, act=None, name=None):
S
sneaxiy 已提交
7637 7638 7639
    return _elementwise_op(LayerHelper('elementwise_sub', **locals()))


X
Xin Pan 已提交
7640
def elementwise_mul(x, y, axis=-1, act=None, name=None):
S
sneaxiy 已提交
7641 7642 7643
    return _elementwise_op(LayerHelper('elementwise_mul', **locals()))


X
Xin Pan 已提交
7644
def elementwise_max(x, y, axis=-1, act=None, name=None):
S
sneaxiy 已提交
7645 7646 7647
    return _elementwise_op(LayerHelper('elementwise_max', **locals()))


X
Xin Pan 已提交
7648
def elementwise_min(x, y, axis=-1, act=None, name=None):
S
sneaxiy 已提交
7649 7650 7651
    return _elementwise_op(LayerHelper('elementwise_min', **locals()))


X
Xin Pan 已提交
7652
def elementwise_pow(x, y, axis=-1, act=None, name=None):
S
sneaxiy 已提交
7653 7654 7655 7656 7657 7658 7659 7660 7661 7662 7663
    return _elementwise_op(LayerHelper('elementwise_pow', **locals()))


for func in [
        elementwise_add, elementwise_div, elementwise_sub, elementwise_mul,
        elementwise_max, elementwise_min, elementwise_pow
]:
    op_proto = OpProtoHolder.instance().get_op_proto(func.__name__)
    func.__doc__ = _generate_doc_string_(
        op_proto,
        additional_args_lines=[
S
sneaxiy 已提交
7664 7665
            "act (basestring|None): Activation applied to the output.",
            "name (basestring|None): Name of the output."
S
sneaxiy 已提交
7666
        ])
M
minqiyang 已提交
7667 7668


7669
def _logical_op(op_name, x, y, out=None, name=None, binary_op=True):
M
minqiyang 已提交
7670 7671
    helper = LayerHelper(op_name, **locals())

M
minqiyang 已提交
7672 7673
    if binary_op:
        assert x.dtype == y.dtype
M
minqiyang 已提交
7674 7675 7676

    if out is None:
        if name is None:
X
Xin Pan 已提交
7677
            out = helper.create_variable_for_type_inference(dtype=x.dtype)
M
minqiyang 已提交
7678 7679 7680 7681 7682 7683 7684 7685 7686 7687 7688 7689 7690 7691 7692
        else:
            out = helper.create_variable(
                name=name, dtype=x.dtype, persistable=False)

    if binary_op:
        helper.append_op(
            type=op_name, inputs={"X": x,
                                  "Y": y}, outputs={"Out": out})
    else:
        helper.append_op(type=op_name, inputs={"X": x}, outputs={"Out": out})

    return out


@templatedoc()
7693
def logical_and(x, y, out=None, name=None):
M
minqiyang 已提交
7694 7695 7696 7697 7698 7699 7700 7701 7702 7703 7704 7705 7706 7707 7708 7709 7710 7711
    """
    ${comment}

    Args:
        x(${x_type}): ${x_comment}
        y(${y_type}): ${y_comment}
        out(Tensor): Output tensor of logical operation.
        name(basestring|None): Name of the output.

    Returns:
        out(${out_type}): ${out_comment}
    """

    return _logical_op(
        op_name="logical_and", x=x, y=y, name=name, out=out, binary_op=True)


@templatedoc()
7712
def logical_or(x, y, out=None, name=None):
M
minqiyang 已提交
7713 7714 7715 7716 7717 7718 7719 7720 7721 7722 7723 7724 7725 7726 7727 7728 7729 7730
    """
    ${comment}

    Args:
        x(${x_type}): ${x_comment}
        y(${y_type}): ${y_comment}
        out(Tensor): Output tensor of logical operation.
        name(basestring|None): Name of the output.

    Returns:
        out(${out_type}): ${out_comment}
    """

    return _logical_op(
        op_name="logical_or", x=x, y=y, name=name, out=out, binary_op=True)


@templatedoc()
7731
def logical_xor(x, y, out=None, name=None):
M
minqiyang 已提交
7732 7733 7734 7735 7736 7737 7738 7739 7740 7741 7742 7743 7744 7745 7746 7747 7748 7749
    """
    ${comment}

    Args:
        x(${x_type}): ${x_comment}
        y(${y_type}): ${y_comment}
        out(Tensor): Output tensor of logical operation.
        name(basestring|None): Name of the output.

    Returns:
        out(${out_type}): ${out_comment}
    """

    return _logical_op(
        op_name="logical_xor", x=x, y=y, name=name, out=out, binary_op=True)


@templatedoc()
7750
def logical_not(x, out=None, name=None):
M
minqiyang 已提交
7751 7752 7753 7754 7755 7756 7757 7758 7759 7760 7761 7762 7763 7764
    """
    ${comment}

    Args:
        x(${x_type}): ${x_comment}
        out(Tensor): Output tensor of logical operation.
        name(basestring|None): Name of the output.

    Returns:
        out(${out_type}): ${out_comment}
    """

    return _logical_op(
        op_name="logical_not", x=x, y=None, name=name, out=out, binary_op=False)
7765 7766 7767 7768 7769 7770 7771 7772 7773 7774 7775 7776 7777 7778 7779 7780 7781 7782 7783 7784


@templatedoc()
def clip(x, min, max, name=None):
    """
    ${comment}

    Args:
        x(${x_type}): ${x_comment}
        min(${min_type}): ${min_comment}
        max(${max_type}): ${max_comment}
        name(basestring|None): Name of the output.

    Returns:
        out(${out_type}): ${out_comment}
    """

    helper = LayerHelper("clip", **locals())

    if name is None:
S
sneaxiy 已提交
7785 7786 7787 7788
        name = unique_name.generate(".".join([helper.name, 'tmp']))

    out = helper.create_variable(
        type=x.type, name=name, dtype=x.dtype, persistable=False)
7789 7790 7791 7792 7793 7794 7795 7796 7797 7798 7799 7800 7801 7802 7803 7804 7805 7806 7807 7808 7809 7810 7811 7812 7813 7814 7815 7816

    helper.append_op(
        type="clip",
        inputs={"X": x},
        attrs={"min": min,
               "max": max},
        outputs={"Out": out})

    return out


@templatedoc()
def clip_by_norm(x, max_norm, name=None):
    """
    ${comment}

    Args:
        x(${x_type}): ${x_comment}
        max_norm(${max_norm_type}): ${max_norm_comment}
        name(basestring|None): Name of the output.

    Returns:
        out(${out_type}): ${out_comment}
    """

    helper = LayerHelper("clip_by_norm", **locals())

    if name is None:
S
sneaxiy 已提交
7817 7818 7819 7820
        name = unique_name.generate(".".join([helper.name, 'tmp']))

    out = helper.create_variable(
        type=x.type, name=name, dtype=x.dtype, persistable=False)
7821 7822 7823 7824 7825 7826 7827 7828

    helper.append_op(
        type="clip_by_norm",
        inputs={"X": x},
        attrs={"max_norm": max_norm},
        outputs={"Out": out})

    return out
X
Xin Pan 已提交
7829 7830 7831 7832 7833 7834 7835 7836 7837 7838 7839 7840 7841 7842 7843 7844 7845 7846


@templatedoc()
def mean(x, name=None):
    """
    ${comment}

    Args:
        x(${x_type}): ${x_comment}
        name(basestring|None): Name of the output.

    Returns:
        out(${out_type}): ${out_comment}
    """

    helper = LayerHelper("mean", **locals())

    if name is None:
X
Xin Pan 已提交
7847
        out = helper.create_variable_for_type_inference(dtype=x.dtype)
X
Xin Pan 已提交
7848 7849 7850 7851 7852 7853 7854 7855 7856 7857 7858 7859 7860 7861 7862 7863 7864 7865 7866 7867 7868 7869 7870 7871 7872 7873 7874 7875 7876
    else:
        out = helper.create_variable(
            name=name, dtype=x.dtype, persistable=False)

    helper.append_op(
        type="mean", inputs={"X": x}, attrs={}, outputs={"Out": out})

    return out


@templatedoc()
def mul(x, y, x_num_col_dims=1, y_num_col_dims=1, name=None):
    """
    ${comment}

    Args:
        x(${x_type}): ${x_comment}
        y(${y_type}): ${y_comment}
        x_num_col_dims(${x_num_col_dims_type}): ${x_num_col_dims_comment}
        y_num_col_dims(${y_num_col_dims_type}): ${y_num_col_dims_comment}
        name(basestring|None): Name of the output.

    Returns:
        out(${out_type}): ${out_comment}
    """

    helper = LayerHelper("mul", **locals())

    if name is None:
X
Xin Pan 已提交
7877
        out = helper.create_variable_for_type_inference(dtype=x.dtype)
X
Xin Pan 已提交
7878 7879 7880 7881 7882 7883 7884 7885 7886
    else:
        out = helper.create_variable(
            name=name, dtype=x.dtype, persistable=False)

    helper.append_op(
        type="mul",
        inputs={"X": x,
                "Y": y},
        attrs={
X
fix  
Xin Pan 已提交
7887 7888
            "x_num_col_dims": x_num_col_dims,
            "y_num_col_dims": y_num_col_dims
X
Xin Pan 已提交
7889 7890 7891 7892 7893 7894
        },
        outputs={"Out": out})
    return out


@templatedoc()
J
jerrywgz 已提交
7895
def sigmoid_cross_entropy_with_logits(x, label, ignore_index=-100, name=None):
X
Xin Pan 已提交
7896 7897 7898 7899 7900 7901
    """
    ${comment}

    Args:
        x(${x_type}): ${x_comment}
        label(${label_type}): ${label_comment}
7902
        ignore_index(&{ignore_index}): ${ignore_index_comment}
X
Xin Pan 已提交
7903 7904 7905 7906 7907 7908 7909 7910 7911
        name(basestring|None): Name of the output.

    Returns:
        out(${out_type}): ${out_comment}
    """

    helper = LayerHelper("sigmoid_cross_entropy_with_logits", **locals())

    if name is None:
X
Xin Pan 已提交
7912
        out = helper.create_variable_for_type_inference(dtype=x.dtype)
X
Xin Pan 已提交
7913 7914 7915 7916 7917 7918 7919 7920
    else:
        out = helper.create_variable(
            name=name, dtype=x.dtype, persistable=False)

    helper.append_op(
        type="sigmoid_cross_entropy_with_logits",
        inputs={"X": x,
                "Label": label},
7921
        attrs={"ignore_index": ignore_index},
X
Xin Pan 已提交
7922 7923 7924 7925 7926 7927 7928 7929 7930 7931 7932 7933 7934 7935 7936 7937 7938 7939 7940 7941
        outputs={"Out": out})
    return out


@templatedoc()
def maxout(x, groups, name=None):
    """
    ${comment}

    Args:
        x(${x_type}): ${x_comment}
        groups(${groups_type}): ${groups_comment}
        name(basestring|None): Name of the output.

    Returns:
        out(${out_type}): ${out_comment}
    """
    helper = LayerHelper("maxout", **locals())

    if name is None:
X
Xin Pan 已提交
7942
        out = helper.create_variable_for_type_inference(dtype=x.dtype)
X
Xin Pan 已提交
7943 7944 7945 7946 7947 7948 7949 7950 7951 7952
    else:
        out = helper.create_variable(
            name=name, dtype=x.dtype, persistable=False)

    helper.append_op(
        type="maxout",
        inputs={"X": x},
        attrs={"groups": groups},
        outputs={"Out": out})
    return out
7953 7954


J
JiabinYang 已提交
7955
def space_to_depth(x, blocksize, name=None):
J
JiabinYang 已提交
7956
    """
J
JiabinYang 已提交
7957
    Gives a blocksize to space_to_depth the input LoDtensor with Layout: [batch, channel, height, width]
J
JiabinYang 已提交
7958
    
J
JiabinYang 已提交
7959
    This op rearranges blocks of spatial data, into depth. More specifically, this op outputs a copy of the 
J
JiabinYang 已提交
7960
    input LoDtensor where values from the height and width dimensions are moved to the channel dimension. 
J
JiabinYang 已提交
7961
    The attr blocksize indicates the input block size.
J
JiabinYang 已提交
7962 7963
    
    space_to_depth will reorgnize the elements of input with shape[batch, channel, height, width] according 
J
JiabinYang 已提交
7964
    to blocksize to construct output with shape [batch, channel * blocksize * blocksize, height/blocksize, width/blocksize]:
J
JiabinYang 已提交
7965 7966 7967
    
    space_to_depth is used to This operation is useful for resizing the activations between convolutions 
    (but keeping all data)
J
JiabinYang 已提交
7968

J
JiabinYang 已提交
7969 7970 7971 7972 7973 7974 7975
    - Non-overlapping blocks of size block_size x block size are rearranged into depth at each location.
    - The depth of the output tensor is block_size * block_size * input channel 
    - The Y, X coordinates within each block of the input become the high order component of the output channel index
    - channel should be divisible by square of blocksize
    - height, width should be divsible by blocksize


J
JiabinYang 已提交
7976
    Args:
J
JiabinYang 已提交
7977
        x(variable): The input LoDtensor.
J
JiabinYang 已提交
7978
        blocksize(variable): The blocksize to select the element on each feature map should be > 2
J
JiabinYang 已提交
7979 7980

    Returns:
J
JiabinYang 已提交
7981
        Variable: The output LoDtensor.
J
JiabinYang 已提交
7982 7983

    Raises:
J
JiabinYang 已提交
7984
        TypeError: blocksize type must be a long.
J
JiabinYang 已提交
7985 7986 7987 7988 7989 7990

    Examples:
        .. code-block:: python

            data = fluid.layers.data(
                name='data', shape=[1, 4, 2, 2], dtype='float32')
J
JiabinYang 已提交
7991
            space_to_depthed = fluid.layers.space_to_depth(
J
JiabinYang 已提交
7992
                x=data, blocksize=2)
J
JiabinYang 已提交
7993 7994
    """

J
JiabinYang 已提交
7995
    helper = LayerHelper("space_to_depth", **locals())
J
JiabinYang 已提交
7996

J
JiabinYang 已提交
7997 7998
    if not (isinstance(blocksize, int)):
        raise ValueError("blocksize must be a python Int")
J
JiabinYang 已提交
7999 8000

    if name is None:
J
JiabinYang 已提交
8001 8002
        out = helper.create_variable_for_type_inference(
            dtype=x.dtype)  #fix create
J
JiabinYang 已提交
8003 8004 8005 8006 8007
    else:
        out = helper.create_variable(
            name=name, dtype=x.dtype, persistable=False)

    helper.append_op(
J
JiabinYang 已提交
8008
        type="space_to_depth",
J
JiabinYang 已提交
8009
        inputs={"X": x},
J
JiabinYang 已提交
8010
        attrs={"blocksize": blocksize},
J
JiabinYang 已提交
8011
        outputs={"Out": out})
J
JiabinYang 已提交
8012 8013
    return out

J
JiabinYang 已提交
8014

S
sneaxiy 已提交
8015 8016 8017 8018 8019 8020 8021 8022 8023 8024 8025 8026 8027 8028
@templatedoc()
def sequence_reverse(x, name=None):
    """ 
    ${comment}

    Args:
        x(${x_type}): ${x_comment}
        name(basestring|None): Name of the output.

    Returns:
        out(${y_type}): ${y_comment}
    """
    helper = LayerHelper("sequence_reverse", **locals())
    if name is None:
S
sneaxiy 已提交
8029
        out = helper.create_variable_for_type_inference(dtype=x.dtype)
S
sneaxiy 已提交
8030 8031 8032 8033 8034 8035 8036 8037 8038 8039
    else:
        out = helper.create_variable(
            name=name, dtype=x.dtype, persistable=False)

    helper.append_op(
        type="sequence_reverse",
        inputs={"X": x},
        outputs={"Y": out},
        attrs=dict())
    return out
S
sneaxiy 已提交
8040 8041


8042 8043 8044 8045 8046 8047
def affine_channel(x, scale=None, bias=None, data_layout='NCHW', name=None):
    """
    Applies a separate affine transformation to each channel of the input.
    Useful for replacing spatial batch norm with its equivalent fixed
    transformation. The input also can be 2D tensor and applies a affine
    transformation in second dimension.
8048

8049 8050 8051 8052 8053 8054 8055 8056 8057 8058 8059 8060 8061 8062 8063 8064 8065 8066 8067
    Args:
        x (Variable): Feature map input can be a 4D tensor with order NCHW
            or NHWC. It also can be a 2D tensor and the affine transformation
            is applied in the second dimension.
        scale (Variable): 1D input of shape (C), the c-th element is the scale
            factor of the affine transformation for the c-th channel of
            the input.
        bias (Variable): 1D input of shape (C), the c-th element is the bias
            of the affine transformation for the c-th channel of the input.
        data_layout (string, default NCHW): NCHW or NHWC. If input is 2D
            tensor, you can ignore data_layout.
        name (str, default None): The name of this layer.

    Returns:
        out (Variable): A tensor of the same shape and data layout with x.
    """
    helper = LayerHelper("affine_channel", **locals())

    if name is None:
X
Xin Pan 已提交
8068
        out = helper.create_variable_for_type_inference(dtype=x.dtype)
8069 8070 8071 8072 8073 8074 8075 8076 8077 8078 8079 8080
    else:
        out = helper.create_variable(
            name=name, dtype=x.dtype, persistable=False)

    helper.append_op(
        type="affine_channel",
        inputs={"X": x,
                'Scale': scale,
                'Bias': bias},
        attrs={"data_layout": data_layout},
        outputs={"Out": out})
    return out
8081 8082


B
barrierye 已提交
8083 8084
def similarity_focus(input, axis, indexes, name=None):
    """  
B
barrierye 已提交
8085
    SimilarityFocus Operator
B
barrierye 已提交
8086 8087

    Generate a similarity focus mask with the same shape of input using the following method:
B
barrierye 已提交
8088
    1. Extract the 3-D tensor(here the first dimension is BatchSize) corresponding 
B
barrierye 已提交
8089
       to the axis according to the indexes. For example, if axis=1 and indexes=[a], 
B
barrierye 已提交
8090
       it will get the matrix T=X[:, a, :, :]. In this case, if the shape of input X 
B
barrierye 已提交
8091 8092 8093 8094
       is (BatchSize, A, B, C), the shape of tensor T is (BatchSize, B, C).
    2. For each index, find the largest numbers in the tensor T, so that the same 
       row and same column has at most one number(what it means is that if the 
       largest number has been found in the i-th row and the j-th column, then 
B
barrierye 已提交
8095 8096 8097
       the numbers in the i-th row or j-th column will be skipped. And then the 
       next largest number will be selected from the remaining numbers. Obviously 
       there will be min(B, C) numbers), and mark the corresponding position of the 
B
barrierye 已提交
8098 8099
       3-D similarity focus mask as 1, otherwise as 0. Do elementwise-or for 
       each index.
B
barrierye 已提交
8100 8101 8102 8103
    3. Broadcast the 3-D similarity focus mask to the same shape of input X.

    Refer to `Similarity Focus Layer <http://www.aclweb.org/anthology/N16-1108>`_

B
barrierye 已提交
8104 8105 8106 8107 8108 8109 8110 8111 8112 8113 8114 8115 8116 8117 8118 8119 8120 8121 8122 8123 8124 8125 8126 8127 8128 8129 8130 8131 8132 8133 8134 8135 8136 8137 8138 8139 8140 8141 8142 8143 8144 8145 8146 8147 8148 8149 8150 8151 8152
    .. code-block:: text

        * Example :

            Given a 4-D tensor x with the shape (BatchSize, C, A, B), where C is
            the number of channels and the shape of feature map is (A, B):
                x.shape = (2, 3, 2, 2)
                x.data = [[[[0.8, 0.1],
                            [0.4, 0.5]],

                           [[0.9, 0.7],
                            [0.9, 0.9]],

                           [[0.8, 0.9],
                            [0.1, 0.2]]],


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

                           [[0.9, 0.7],
                            [0.8, 0.4]],

                           [[0.0, 0.2],
                            [0.4, 0.7]]]]

            Given axis: 1 (the axis of the channel)
            Given indexes: [0]

            then we get a 4-D tensor out with the same shape of input x:
                out.shape = (2, 3, 2, 2)
                out.data = [[[[1.0, 0.0],
                              [0.0, 1.0]],

                             [[1.0, 0.0],
                              [0.0, 1.0]],

                             [[1.0, 0.0],
                              [0.0, 1.0]]],

                            [[[0.0, 1.0],
                              [1.0, 0.0]],

                             [[0.0, 1.0],
                              [1.0, 0.0]],

                             [[0.0, 1.0],
                              [1.0, 0.0]]]]

B
barrierye 已提交
8153 8154 8155
    Args:
        input(Variable): The input tensor variable(default float). It should 
            be a 4-D tensor with shape [BatchSize, A, B, C].
B
barrierye 已提交
8156
        axis(int): Indicating the dimension to be selected. It can only be
B
barrierye 已提交
8157
            1, 2 or 3.
B
barrierye 已提交
8158
        indexes(list): Indicating the indexes of the selected dimension.
B
barrierye 已提交
8159 8160 8161 8162 8163 8164 8165 8166

    Returns:
        Variable: A tensor variable with the same shape and same type 
            as the input.
        
    Examples:
        .. code-block:: python
            data = fluid.layers.data(
B
barrierye 已提交
8167 8168
              name='data', shape=[2, 3, 2, 2], dtype='float32')
            x = fluid.layers.layer_norm(input=data, axis=1, indexes=[0])
B
barrierye 已提交
8169 8170 8171 8172 8173 8174 8175 8176 8177 8178 8179 8180
    """
    helper = LayerHelper('similarity_focus', **locals())
    # check attrs
    if isinstance(axis, int) is False:
        raise TypeError("axis must be int type.")
    if isinstance(indexes, list) is False:
        raise TypeError("indexes must be list type.")
    if axis != 1 and axis != 2 and axis != 3:
        raise ValueError("axis must be 1, 2 or 3.")
    if len(indexes) == 0:
        raise ValueError("indexes can not be empty.")

B
barrierye 已提交
8181 8182 8183 8184 8185
    if name is None:
        out = helper.create_variable_for_type_inference(dtype=input.dtype)
    else:
        out = helper.create_variable(
            name=name, dtype=input.dtype, persistable=False)
B
barrierye 已提交
8186 8187 8188 8189 8190 8191 8192
    helper.append_op(
        type='similarity_focus',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={"axis": axis,
               "indexes": indexes})
    return out
B
barrierye 已提交
8193 8194


M
minqiyang 已提交
8195 8196
def hash(input, hash_size, num_hash=1, name=None):
    """
M
minqiyang 已提交
8197 8198
    Hash the input to an integer whose value is less than the given hash size.

M
minqiyang 已提交
8199 8200
    The hash algorithm we used was xxHash - Extremely fast hash algorithm
    (https://github.com/Cyan4973/xxHash/tree/v0.6.5)
M
minqiyang 已提交
8201 8202 8203 8204 8205 8206 8207 8208 8209 8210 8211 8212 8213 8214 8215 8216 8217 8218 8219 8220 8221 8222 8223 8224 8225 8226 8227 8228 8229 8230 8231 8232 8233 8234 8235 8236 8237 8238

    A simple example as below:

    .. code-block:: text

        Given:

        # shape [2, 2]
        input.data = [
            [[1], [2]],
            [[3], [4]],
        ]

        input.lod = [[0, 2]]

        hash_size = 10000

        num_hash = 4

        Then:

        Hash op will take all number in input's 2nd dimension as hash algorithm's
        input for each time. Each input will be hashed for 4 times, and get an
        array whose length is 4. Each value in the array ranges from 0 to 9999.

        # shape [2, 4]
        output.data = [
            [[9662], [9217], [1129], [8487]],
            [[8310], [1327], [1654], [4567]],
        ]

        output.lod = [[0, 2]]

    Args:
        input (Variable): The input variable which is a one-hot word. The
            dimensions of the input variable must be 2.
        hash_size (int): The space size for hash algorithm. The output value
            will keep in the range:math:`[0, hash_size - 1]`.
M
minqiyang 已提交
8239
        num_hash (int): The times of hash, default 1.
M
minqiyang 已提交
8240
        name (str, default None): The name of this layer.
M
minqiyang 已提交
8241 8242 8243 8244 8245 8246 8247 8248 8249

    Returns:
       Variable: The hash result variable which is a LoDTensor.

    Examples:
       .. code-block:: python
           word_dict = paddle.dataset.imdb.word_dict()
           x = fluid.layers.data(shape[1], dtype='int32', lod_level=1)
           out = fluid.layers.hash(input=x, num_hash=4, hash_size=1000)
M
minqiyang 已提交
8250 8251
    """
    helper = LayerHelper('hash', **locals())
M
minqiyang 已提交
8252 8253
    out = helper.create_variable_for_type_inference(
        helper.input_dtype(), stop_gradient=True)
M
minqiyang 已提交
8254 8255 8256 8257 8258 8259 8260
    helper.append_op(
        type='hash',
        inputs={'X': input},
        outputs={'Out': out},
        attrs={'num_hash': num_hash,
               'mod_by': hash_size})
    return out
G
gmcather 已提交
8261 8262


D
dengkaipeng 已提交
8263
@templatedoc()
8264 8265
def grid_sampler(x, grid, name=None):
    """
8266 8267 8268 8269 8270 8271 8272
    This operation samples input X by using bilinear interpolation based on 
    flow field grid, which is usually gennerated by affine_grid. The grid of
    shape [N, H, W, 2] is the concatenation of (grid_x, grid_y) coordinates 
    with shape [N, H, W] each, where grid_x is indexing the 4th dimension 
    (in width dimension) of input data x and grid_y is indexng the 3rd 
    dimention (in height dimension), finally results is the bilinear 
    interpolation value of 4 nearest corner points.
8273 8274 8275 8276 8277 8278 8279 8280 8281 8282 8283 8284 8285 8286 8287 8288 8289 8290 8291 8292 8293 8294 8295 8296 8297 8298 8299 8300 8301 8302 8303 8304 8305 8306 8307 8308 8309 8310

    Step 1:
    Get (x, y) grid coordinates and scale to [0, H-1/W-1].

    grid_x = 0.5 * (grid[:, :, :, 0] + 1) * (W - 1)
    grid_y = 0.5 * (grid[:, :, :, 1] + 1) * (H - 1)

    Step 2:
    Indices input data X with grid (x, y) in each [H, W] area, and bilinear 
    interpolate point value by 4 nearest points.

      wn ------- y_n ------- en
      |           |           |
      |          d_n          |
      |           |           |
     x_w --d_w-- grid--d_e-- x_e
      |           |           |
      |          d_s          |
      |           |           |
      ws ------- y_s ------- wn

    x_w = floor(x)              // west side x coord
    x_e = x_w + 1               // east side x coord
    y_n = floor(y)              // north side y coord
    y_s = y_s + 1               // south side y coord

    d_w = grid_x - x_w          // distance to west side
    d_e = x_e - grid_x          // distance to east side
    d_n = grid_y - y_n          // distance to north side
    d_s = y_s - grid_y          // distance to south side

    wn = X[:, :, y_n, x_w]      // north-west point value
    en = X[:, :, y_n, x_e]      // north-east point value
    ws = X[:, :, y_s, x_w]      // south-east point value
    es = X[:, :, y_s, x_w]      // north-east point value

    output = wn * d_e * d_s + en * d_w * d_s
           + ws * d_e * d_n + es * d_w * d_n
D
dengkaipeng 已提交
8311 8312

    Args:
8313 8314 8315
        x(Variable): Input data of shape [N, C, H, W].
        grid(Variable): Input grid tensor of shape [N, H, W, 2].
        name (str, default None): The name of this layer.
D
dengkaipeng 已提交
8316 8317

    Returns:
8318 8319 8320 8321 8322 8323 8324 8325 8326 8327
        out(Variable): Output of shape [N, C, H, W] data samples input X 
        using bilnear interpolation based on input grid.

    Exmples:
    .. code-block:: python

        x = fluid.layers.data(name='x', shape=[3, 10, 32, 32], dtype='float32')
        theta = fluid.layers.data(name='theta', shape=[3, 2, 3], dtype='float32')
        grid = fluid.layers.affine_grid(input=theta, size=[3, 10, 32, 32]})
        out = fluid.layers.grid_sampler(x=x, grid=grid)
D
dengkaipeng 已提交
8328 8329 8330 8331 8332 8333 8334 8335 8336
    """
    helper = LayerHelper("grid_sampler", **locals())

    if not isinstance(x, Variable):
        return ValueError("The x should be a Variable")

    if not isinstance(grid, Variable):
        return ValueError("The grid should be a Variable")

8337
    out = helper.create_variable_for_type_inference(x.dtype)
D
dengkaipeng 已提交
8338 8339
    ipts = {'X': x, 'Grid': grid}

8340
    helper.append_op(type='grid_sampler', inputs=ipts, outputs={'Output': out})
8341 8342 8343
    return out


G
gmcather 已提交
8344 8345 8346 8347 8348 8349 8350 8351 8352 8353 8354 8355 8356 8357 8358 8359 8360 8361 8362 8363 8364 8365 8366 8367 8368 8369 8370 8371 8372 8373 8374 8375 8376 8377 8378 8379 8380 8381 8382 8383 8384 8385 8386 8387 8388 8389 8390 8391 8392 8393 8394 8395 8396 8397 8398 8399 8400 8401 8402 8403 8404 8405 8406 8407 8408 8409 8410 8411 8412 8413 8414 8415 8416 8417 8418 8419 8420 8421 8422 8423 8424 8425 8426 8427 8428 8429 8430 8431 8432 8433 8434 8435 8436 8437
def log_loss(input, label, epsilon=1e-4, name=None):
    """
    **Negative Log Loss Layer**

    This layer accepts input predictions and target label and returns the
    negative log loss.

    .. math::

        Out = -label * \\log{(input + \\epsilon)}
              - (1 - label) * \\log{(1 - input + \\epsilon)}

    Args:
        input (Variable|list):  a 2-D tensor with shape [N x 1], where N is the
                                batch size. This input is a probability computed
                                by the previous operator.
        label (Variable|list):  the ground truth which is a 2-D tensor with
                                shape [N x 1], where N is the batch size.
        epsilon (float): epsilon
        name (string): the name of log_loss

    Returns:
        Variable: A 2-D tensor with shape [N x 1], the negative log loss.

    Examples:
        .. code-block:: python

          prob = fluid.layers.sigmoid(net)
          cost = fluid.layers.log_loss(input=prob, label=label)
    """
    helper = LayerHelper('log_loss', **locals())

    if name is None:
        loss = helper.create_variable_for_type_inference(dtype=input.dtype)
    else:
        loss = helper.create_variable(
            name=name, dtype=input.dtype, persistable=False)

    helper.append_op(
        type='log_loss',
        inputs={'Predicted': [input],
                'Labels': [label]},
        outputs={'Loss': [loss]},
        attrs={'epsilon': epsilon})
    return loss


def add_position_encoding(input, alpha, beta, name=None):
    """
    **Add Position Encoding Layer**

    This layer accepts an input 3D-Tensor of shape [N x M x P], and return an
    output Tensor of shape [N x M x P] with positional encoding value.

    Refer to `Attention Is All You Need<http://arxiv.org/pdf/1706.03762.pdf>`_ .

    .. math::
        PE(pos, 2i) = \\sin{(pos / 10000^{2i / P})}   \\\\
        PE(pos, 2i + 1) = \\cos{(pos / 10000^{2i / P})}  \\\\
        Out(:, pos, i) = \\alpha * input(:, pos, i) + \\beta * PE(pos, i)

    Where:
    * PE(pos, 2i): the increment for the number at even position
    * PE(pos, 2i + 1): the increment for the number at odd position

    Args:
        input (Variable): 3-D input tensor with shape [N x M x P]
        alpha (float): multiple of Input Tensor
        beta (float): multiple of Positional Encoding Tensor
        name (string): the name of position encoding layer

    Returns:
        Variable: A 3-D Tensor of shape [N x M x P] with positional encoding.

    Examples:
        .. code-block:: python

          position_tensor = fluid.layers.add_position_encoding(input=tensor)
    """
    helper = LayerHelper('add_position_encoding', **locals())
    dtype = helper.input_dtype()

    if name is None:
        out = helper.create_variable_for_type_inference(dtype=dtype)
    else:
        out = helper.create_variable(name=name, dtype=dtype, persistable=False)

    helper.append_op(
        type="add_position_encoding",
        inputs={"X": input},
        outputs={"Out": out},
        attrs={"alpha": alpha,
               "beta": beta})
    return out
Q
Qiao Longfei 已提交
8438 8439 8440 8441 8442 8443 8444 8445 8446 8447


def bilinear_tensor_product(x,
                            y,
                            size,
                            act=None,
                            name=None,
                            param_attr=None,
                            bias_attr=None):
    """
Q
Qiao Longfei 已提交
8448
    **Add Bilinear Tensor Product Layer**
Q
Qiao Longfei 已提交
8449

Q
Qiao Longfei 已提交
8450
    This layer performs bilinear tensor product on two inputs.
Q
Qiao Longfei 已提交
8451 8452 8453
    For example:

    .. math::
8454
       out{i} = x * W_{i} * {y^\mathrm{T}}, i=0,1,...,size-1
Q
Qiao Longfei 已提交
8455

Q
Qiao Longfei 已提交
8456
    In this formula:
8457 8458
      - :math:`x`: the first input contains M elements, shape is [batch_size, M].
      - :math:`y`: the second input contains N elements, shape is [batch_size, N].
Q
Qiao Longfei 已提交
8459
      - :math:`W_{i}`: the i-th learned weight, shape is [M, N]
8460
      - :math:`out{i}`: the i-th element of out, shape is [batch_size, size].
Q
Qiao Longfei 已提交
8461 8462 8463
      - :math:`y^\mathrm{T}`: the transpose of :math:`y_{2}`.

    Args:
8464 8465
        x (Variable): 2-D input tensor with shape [batch_size, M]
        y (Variable): 2-D input tensor with shape [batch_size, N]
Q
Qiao Longfei 已提交
8466 8467 8468
        size (int): The dimension of this layer.
        act (str, default None): Activation to be applied to the output of this layer.
        name (str, default None): The name of this layer.
Q
Qiao Longfei 已提交
8469
        param_attr (ParamAttr, default None): The parameter attribute for the learnable w.
Q
Qiao Longfei 已提交
8470
            parameters/weights of this layer.
Q
Qiao Longfei 已提交
8471
        bias_attr (ParamAttr, default None): The parameter attribute for the bias
Q
Qiao Longfei 已提交
8472 8473 8474 8475
            of this layer. If it is set to False, no bias will be added to the output units.
            If it is set to None, the bias is initialized zero. Default: None.

    Returns:
Q
Qiao Longfei 已提交
8476
        Variable: A 2-D Tensor of shape [batch_size, size].
Q
Qiao Longfei 已提交
8477 8478 8479 8480

    Examples:
        .. code-block:: python

Q
Qiao Longfei 已提交
8481
          tensor = bilinear_tensor_product(x=layer1, y=layer2, size=1000)
Q
Qiao Longfei 已提交
8482 8483
    """
    helper = LayerHelper('bilinear_tensor_product', **locals())
Q
Qiao Longfei 已提交
8484
    dtype = helper.input_dtype('x')
Q
Qiao Longfei 已提交
8485 8486 8487 8488

    param_shape = [size, x.shape[1], y.shape[1]]

    w = helper.create_parameter(
Q
Qiao Longfei 已提交
8489
        attr=helper.param_attr, shape=param_shape, dtype=dtype, is_bias=False)
Q
Qiao Longfei 已提交
8490 8491 8492 8493 8494 8495 8496 8497 8498 8499 8500 8501 8502 8503 8504 8505 8506

    if name is None:
        out = helper.create_variable_for_type_inference(dtype=dtype)
    else:
        out = helper.create_variable(name=name, dtype=dtype, persistable=False)

    inputs = {"X": x, "Y": y, "Weight": w}
    if helper.bias_attr:
        bias_size = [1, size]
        bias = helper.create_parameter(
            attr=helper.bias_attr, shape=bias_size, dtype=dtype, is_bias=True)
        inputs["Bias"] = bias
    helper.append_op(
        type="bilinear_tensor_product", inputs=inputs, outputs={"Out": out})

    # add activation
    return helper.append_activation(out)