config_parser.py 153.7 KB
Newer Older
1
# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved
Z
zhangjinchao01 已提交
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 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 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 99 100 101
#
# 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.

from __future__ import print_function
'''
The following functions are available in the config file:

Bias: define bias. To be used as value of bias argument in Layer().

Data: define data provider.

Input: define input layer for a layer. To be used as element of inputs argument
       in Layer().

Conv: define a convolution operation for an input of a layer.

Norm: define a normalization operation for an input of a layer.

Pool: define a pooling operation for an input of a layer.

Layer: define a layer.

Parameter: define a parameter.

Import: import another config file. If the imported config file name is
        a relative path, then it will be searched under the directory of the
        current config file.

Inputs(layer_names...):
    Define the name of the input layers of the NeuralNetwork.
    The type of these layers must be "data".
    These layers will be provided with the DataBatch obtained
    from DataProvider. The data streams from DataProvider must
    have the same order.

Outputs(layer_names...):
    Define the name of the output layers of the NeuralNetwork.
    Usually the output is simply the cost layer.
    You can specify other layers as outputs and  calculate the
    cost (and its derivative) yourself.


default_initial_std(val)
default_initial_mean(val)
default_momentum(val):
default_decay_rate(val): Set the default value for these parameters


get_config_arg(name, type, default): Get the value for a config parameter.


*** customized extension to config_parser ***
The functionality of the config_parser can be extended.
If the config_arg_str for parse_config() contains
extension_module_name=[MODULE_NAME], then config_parser will call
MODULE_NAME.get_config_funcs(g_config)
MODULE_NAME.get_config_funcs() should return a dictionary of name to functions,
those functions will be available in the config file.
See trainer/tests/config_parser_test.py for example

To use this from paddle_trainer, paddle_trainer should be called with
--config_args=extension_module_name=[MODULE_NAME]

'''
import copy
import logging
import os
import sys
import traceback
import math
import shutil

try:
    from paddle.proto.DataConfig_pb2 import DataConfig
    from paddle.proto.ModelConfig_pb2 import ModelConfig
    from paddle.proto.ModelConfig_pb2 import LayerConfig
    from paddle.proto.ModelConfig_pb2 import LayerInputConfig
    from paddle.proto.ModelConfig_pb2 import ProjectionConfig
    from paddle.proto.ModelConfig_pb2 import OperatorConfig
    from paddle.proto.ModelConfig_pb2 import GeneratorConfig
    from paddle.proto.ModelConfig_pb2 import LinkConfig
    from paddle.proto.ParameterConfig_pb2 import ParameterConfig
    from paddle.proto.ParameterConfig_pb2 import ParameterUpdaterHookConfig
    from paddle.proto.TrainerConfig_pb2 import TrainerConfig

except Exception as e:
    traceback.print_exc()
    raise

logging.basicConfig(
Q
qijun 已提交
102
    format='[%(levelname)s %(asctime)s %(filename)s:%(lineno)s] %(message)s', )
Z
zhangjinchao01 已提交
103 104 105
logger = logging.getLogger('paddle')
logger.setLevel(logging.INFO)
__real_print__ = print
Q
qijun 已提交
106
print = logger.info
Z
zhangjinchao01 已提交
107 108 109 110

# from layer type name to layer class
g_layer_type_map = {}

Q
qijun 已提交
111

Z
zhangjinchao01 已提交
112 113 114
# Initialize global variables. We use this function so that we can
# call parse_config() multiple times
def init_config_environment(
Q
qijun 已提交
115 116 117 118 119 120 121 122 123 124 125 126 127 128
        g_default_momentum=None,
        g_default_decay_rate=None,
        g_default_initial_mean=0.,
        g_default_initial_std=0.01,
        g_default_num_batches_regularization=None,
        g_default_initial_strategy=0,
        g_default_initial_smart=False,
        g_default_gradient_clipping_threshold=None,
        g_default_device=None,
        g_default_update_hooks=None,
        g_default_compact_func=None,
        g_config=TrainerConfig(),
        g_layer_map={},
        g_parameter_map={},
X
xuwei06 已提交
129
        g_parameter_initializer_map={},
Q
qijun 已提交
130
        g_extended_config_funcs={},
Z
zhangjinchao01 已提交
131 132

        # store command args of paddle_trainer
Q
qijun 已提交
133
        g_command_config_args={},
Z
zhangjinchao01 已提交
134 135

        # Used for PyDataProvider to avoid duplicate module name
Q
qijun 已提交
136 137 138 139 140
        g_py_module_name_list=[],
        g_current_submodel=None,
        g_root_submodel=None,
        g_submodel_map={},
        g_submodel_stack=[],
141
        g_add_submodel_suffix=False, ):
Z
zhangjinchao01 已提交
142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157

    for k, v in locals().iteritems():
        globals()[k] = copy.deepcopy(v)


# Because type is widely used as a variable name in this code.
# we need a different function name for the builtin type()
def type_of(x):
    return type(x)


# Check a condition derived config file
def config_assert(b, msg):
    if not b:
        logger.fatal(msg)

Q
qijun 已提交
158

Z
zhangjinchao01 已提交
159 160
g_config_funcs = {}

Q
qijun 已提交
161

Z
zhangjinchao01 已提交
162 163 164 165 166
# decorator for indicating a function which can be used in config file
def config_func(func):
    g_config_funcs[func.func_name] = func
    return func

Q
qijun 已提交
167

Z
zhangjinchao01 已提交
168 169 170 171 172
# decorator for indicating a class which can be used in config file
def config_class(cls):
    g_config_funcs[cls.__name__] = cls
    return cls

Q
qijun 已提交
173

Z
zhangjinchao01 已提交
174 175 176 177 178 179
# decorator for indicating a class for a layer type
def config_layer(layer_type):
    def wrap(cls):
        g_config_funcs[cls.__name__] = cls
        g_layer_type_map[layer_type] = cls
        return cls
Q
qijun 已提交
180

Z
zhangjinchao01 已提交
181 182
    return wrap

Q
qijun 已提交
183

Z
zhangjinchao01 已提交
184 185 186
def gen_parameter_name(layer_name, input_index):
    return '_%s.w%d' % (layer_name, input_index)

Q
qijun 已提交
187

Z
zhangjinchao01 已提交
188 189 190
def gen_bias_parameter_name(layer_name):
    return '_%s.wbias' % layer_name

Q
qijun 已提交
191

Z
zhangjinchao01 已提交
192 193 194
def default(x, default_value):
    return default_value if x is None else x

Q
qijun 已提交
195

Z
zhangjinchao01 已提交
196 197 198 199 200 201
class Cfg(object):
    def add_keys(self, locals):
        for k, v in locals.iteritems():
            if not k.startswith('_'):
                self.__setattr__(k, v)

Q
qijun 已提交
202

Z
zhangjinchao01 已提交
203 204
# functions available in config file

Q
qijun 已提交
205

Z
zhangjinchao01 已提交
206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223
# Define the name of the input layers of the NeuralNetwork.
# The type of these layers must be "data".
# These layers will be provided with the DataBatch obtained
# from DataProvider. The data streams from DataProvider must
# have the same order.
@config_func
def Inputs(*args):
    for name in args:
        name = MakeLayerNameInSubmodel(name)
        global g_current_submodel, g_root_submodel
        if g_current_submodel.is_recurrent_layer_group:
            config_assert(False, "Do not set Inputs in recurrent layer group")
        else:
            g_current_submodel.input_layer_names.append(name)

        if g_current_submodel is g_root_submodel:
            g_config.model_config.input_layer_names.append(name)

Q
qijun 已提交
224

225 226
@config_func
def HasInputsSet():
227
    return len(g_current_submodel.input_layer_names) != 0
228

Z
zhangjinchao01 已提交
229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252

# Define the name of the output layers of the NeuralNetwork.
# Usually the output is simply the cost layer.
# You can specify other layers as outputs and calculate the
# cost (and its derivative) yourself.
@config_func
def Outputs(*args):
    for name in args:
        name = MakeLayerNameInSubmodel(name)
        global g_current_submodel, g_root_submodel
        if g_current_submodel.is_recurrent_layer_group:
            config_assert(False, "Do not set Outputs in recurrent layer group")
        else:
            g_current_submodel.output_layer_names.append(name)

        if g_current_submodel is g_root_submodel:
            g_config.model_config.output_layer_names.append(name)


@config_func
def SubModelBegin(name):
    global g_current_submodel, g_root_submodel, g_submodel_stack
    g_submodel_stack.append(g_current_submodel)

Q
qijun 已提交
253
    name = MakeLayerNameInParentSubmodel(name)  #rename in nested submodel
Z
zhangjinchao01 已提交
254 255 256 257 258 259 260 261 262

    config_assert(name not in g_submodel_map,
                  'Duplicated submodel name: %s' % name)

    sub_model = g_config.model_config.sub_models.add()
    sub_model.name = name
    g_submodel_map[name] = sub_model
    g_current_submodel = sub_model

Q
qijun 已提交
263

Z
zhangjinchao01 已提交
264
@config_func
Q
qijun 已提交
265
def SubModelEnd(name=None):
Z
zhangjinchao01 已提交
266
    global g_current_submodel, g_root_submodel, g_submodel_stack
Q
qijun 已提交
267 268
    config_assert(g_current_submodel is not g_root_submodel,
                  "submodel not begin")
Z
zhangjinchao01 已提交
269
    if name is not None:
Q
qijun 已提交
270 271 272
        config_assert(
            g_current_submodel.name == MakeLayerNameInParentSubmodel(name),
            "submodel name error")
Z
zhangjinchao01 已提交
273 274 275

    g_current_submodel = g_submodel_stack.pop()

Q
qijun 已提交
276

Z
zhangjinchao01 已提交
277 278
def MakeLayerNameInParentSubmodel(name):
    suffix = ""
279 280
    if len(g_submodel_stack) > 1:
        suffix = "@" + g_submodel_stack[-1].name
Z
zhangjinchao01 已提交
281 282
    return name + suffix

Q
qijun 已提交
283

Z
zhangjinchao01 已提交
284 285 286
def GetLayerBaseName(name):
    return name.split('@')[0]

Q
qijun 已提交
287 288

def MakeLayerNameInSubmodel(name, submodel_name=None):
Z
zhangjinchao01 已提交
289 290
    global g_current_submodel
    global g_add_submodel_suffix
Q
qijun 已提交
291 292
    if (submodel_name is None and not g_add_submodel_suffix and
            not g_current_submodel.is_recurrent_layer_group):
Z
zhangjinchao01 已提交
293 294 295 296 297
        return name
    if submodel_name is None:
        submodel_name = g_current_submodel.name
    return name + "@" + submodel_name

Q
qijun 已提交
298

Z
zhangjinchao01 已提交
299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321
# Define a recurrent layer group begin with RecurrentLayerGroupBegin
# and end with RecurrentLayerGroupEnd.
# A recurrent layer group forward/backward one frame after previous frame
# forward/backward through all layers in layer group.
# in_links are names of layer used as input layer in the layer group.
# out_links are names of layer in layer group used as outside layer's input.
#
# If generator is set, the layer group need one or more than one outlinks.
# The first outlink should always be the generated token ids.
# If generator.num_results_per_sample is not set, the output for one sample is
# a ids sequence. Else if num_results_per_sample is more than one,
# the output for one sample is up to #num_results_per_sample generated
# sequences, which are packed in one sequence in output ids vector. Each
# generated sequence has a generation probability. The probabilities for one
# sample are stored in one row of output value matrix.
# Packed generated sequences format, for each i:
#   seq_i_length: one interger, seq_i content length,
#   [seq_i content], length = seq_i_length
#   seq_i_end_mark: one interger, for format check, always -1
# You can use "seq_text_printer" to print the output of the generator.
@config_func
def RecurrentLayerGroupWithoutOutLinksBegin(name,
                                            in_links,
322 323
                                            seq_reversed=False,
                                            target_inlinkname=""):
Z
zhangjinchao01 已提交
324 325 326 327 328 329 330 331
    global g_current_submodel
    config_assert(g_config.model_config.type == "recurrent_nn",
                  "RecurrentLayerGroup should be used only in recurrent_nn")
    RecurrentLayerGroup(name=name)  # add to father model
    SubModelBegin(name)
    g_current_submodel.is_recurrent_layer_group = True
    g_current_submodel.reversed = seq_reversed
    in_links_count = 0
332
    for linkid, link in enumerate(in_links):
Z
zhangjinchao01 已提交
333 334 335 336
        if isinstance(link, basestring):
            name = link
        else:
            name = link.link_name
337

Z
zhangjinchao01 已提交
338 339 340
        in_links_count += 1
        layer_name = MakeLayerNameInParentSubmodel(name)
        layer = g_layer_map[layer_name]
341 342
        ScatterAgentLayer(
            name=name, size=layer.size, width=layer.width, height=layer.height)
343

Z
zhangjinchao01 已提交
344 345 346 347
        pair = g_current_submodel.in_links.add()
        pair.layer_name = layer_name
        pair.link_name = MakeLayerNameInSubmodel(name)

Q
qijun 已提交
348

Z
zhangjinchao01 已提交
349 350 351 352 353 354 355 356 357 358 359 360 361
@config_func
def RecurrentLayerGroupSetOutLink(link):
    if isinstance(link, basestring):
        name = link
    else:
        name = link.link_name
    layer_name = MakeLayerNameInParentSubmodel(name)
    pair = g_current_submodel.out_links.add()
    pair.layer_name = MakeLayerNameInSubmodel(name)
    pair.link_name = layer_name


def RecurrentLayerGroupSetGenerator(generator=None):
Q
qijun 已提交
362
    generator.eos_layer_name = MakeLayerNameInSubmodel(generator.eos_layer_name)
Z
zhangjinchao01 已提交
363 364 365 366 367 368 369 370
    g_current_submodel.generator.CopyFrom(generator)


@config_func
def RecurrentLayerGroupBegin(name,
                             in_links,
                             out_links,
                             generator=None,
371
                             target_inlinkname="",
Z
zhangjinchao01 已提交
372
                             seq_reversed=False):
373
    RecurrentLayerGroupWithoutOutLinksBegin(name, in_links, seq_reversed)
Z
zhangjinchao01 已提交
374 375 376 377 378
    for link in out_links:
        RecurrentLayerGroupSetOutLink(link)

    if generator is not None:
        RecurrentLayerGroupSetGenerator(generator)
Q
qijun 已提交
379 380 381 382 383
        config_assert(
            len(in_links) == 0, "no in_links should be passed to generator")
        config_assert(
            len(out_links) >= 1,
            "one or more than one out_links should be passed to generator")
Z
zhangjinchao01 已提交
384 385 386 387 388 389 390


@config_func
def RecurrentLayerGroupEnd(name):
    global g_current_submodel
    config_assert(g_current_submodel.is_recurrent_layer_group,
                  "RecurrentLayerGroup not begin")
Q
qijun 已提交
391
    for pair in g_current_submodel.memories:  #check exist
Z
zhangjinchao01 已提交
392
        layer = g_layer_map[pair.layer_name]
Y
Yu Yang 已提交
393 394
        config_assert(layer is not None,
                      "memory declare wrong name:%s" % pair.layer_name)
Z
zhangjinchao01 已提交
395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410
        memory_link = g_layer_map[pair.link_name]
        config_assert(layer.size == memory_link.size,
                      "memory declare wrong size:%d" % memory_link.size)

    prev_submodel = g_current_submodel
    SubModelEnd(name)

    for pair in prev_submodel.out_links:
        layer = g_layer_map[pair.layer_name]
        # add out agent to father model
        agent_name = GetLayerBaseName(pair.link_name)
        if prev_submodel.HasField("generator"):
            DataLayer(name=agent_name, size=layer.size)
        else:
            GatherAgentLayer(name=agent_name, size=layer.size)

Q
qijun 已提交
411

Z
zhangjinchao01 已提交
412 413 414 415 416 417
# Define the model type
# currently, the paddle supports "nn", "recurrent_nn", "recursive_nn" and "multi_nn"
@config_func
def model_type(name):
    g_config.model_config.type = name

Q
qijun 已提交
418

Z
zhangjinchao01 已提交
419 420
@config_class
class Bias(Cfg):
X
xuwei06 已提交
421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436
    def __init__(self,
                 parameter_name=None,
                 learning_rate=None,
                 momentum=None,
                 decay_rate=None,
                 decay_rate_l1=None,
                 initial_mean=None,
                 initial_std=None,
                 initial_strategy=None,
                 initial_smart=None,
                 num_batches_regularization=None,
                 sparse_remote_update=None,
                 gradient_clipping_threshold=None,
                 is_static=None,
                 is_shared=None,
                 initializer=None):
Z
zhangjinchao01 已提交
437 438
        self.add_keys(locals())

Q
qijun 已提交
439

Z
zhangjinchao01 已提交
440 441 442 443 444 445 446
# Define one input for a layer
@config_class
class Input(Cfg):
    def __init__(
            self,
            input_layer_name,
            parameter_name=None,
X
xuwei06 已提交
447
            initializer=None,
Z
zhangjinchao01 已提交
448 449 450 451 452 453 454 455 456 457 458 459 460
            learning_rate=None,
            momentum=None,
            decay_rate=None,
            decay_rate_l1=None,
            initial_mean=None,
            initial_std=None,
            initial_strategy=None,
            initial_smart=None,
            num_batches_regularization=None,
            sparse_remote_update=None,
            sparse_update=None,
            gradient_clipping_threshold=None,
            conv=None,
L
liaogang 已提交
461
            bilinear_interp=None,
Z
zhangjinchao01 已提交
462 463 464 465
            norm=None,
            pool=None,
            image=None,
            block_expand=None,
466
            maxout=None,
Q
qijun 已提交
467
            spp=None,
D
dangqingqing 已提交
468
            pad=None,
Z
zhangjinchao01 已提交
469 470 471 472 473
            format=None,
            nnz=None,
            is_static=None,
            is_shared=None,
            update_hooks=None,
474
            input_layer_argument=None,
D
dangqingqing 已提交
475 476 477 478 479
            make_layer_name_in_submodel=True, ):
        """
        @param make_layer_name_in_submodel True by defalut, you might need to
        set it carefully when adding Input in config_parser.py.
        """
Z
zhangjinchao01 已提交
480
        self.add_keys(locals())
D
dangqingqing 已提交
481 482 483
        self.input_layer_name = MakeLayerNameInSubmodel(
            input_layer_name
        ) if make_layer_name_in_submodel else input_layer_name
Z
zhangjinchao01 已提交
484

Q
qijun 已提交
485

Z
zhangjinchao01 已提交
486 487 488
# Define a projection for iexed layer
@config_class
class Projection(Input):
Q
qijun 已提交
489 490
    type = None  # subclass should set it correctly

Z
zhangjinchao01 已提交
491 492 493
    def __init__(
            self,
            input_layer_name,
Q
qijun 已提交
494
            size=0,  # projection output size
Z
zhangjinchao01 已提交
495 496 497 498 499 500 501 502 503
            parameter_name=None,
            learning_rate=None,
            momentum=None,
            decay_rate=None,
            decay_rate_l1=None,
            initial_mean=None,
            initial_std=None,
            initial_strategy=None,
            initial_smart=None,
X
xuwei06 已提交
504
            initializer=None,
Z
zhangjinchao01 已提交
505 506 507 508 509 510 511 512 513 514
            num_batches_regularization=None,
            sparse_remote_update=None,
            sparse_update=None,
            gradient_clipping_threshold=None,
            ptype=None,
            format=None,
            nnz=None,
            is_static=None,
            is_shared=None,
            update_hooks=None,
Q
qijun 已提交
515
            input_layer_argument=None, ):
Z
zhangjinchao01 已提交
516 517 518 519 520 521 522 523 524 525 526 527 528
        self.add_keys(locals())
        self.input_layer_name = MakeLayerNameInSubmodel(input_layer_name)

        self.proj_conf = ProjectionConfig()
        if ptype is not None:
            self.proj_conf.type = ptype
        else:
            self.proj_conf.type = self.type

    # calculate the output_size given input_size. return 0
    # to indicate using the size from Layer config
    def calc_output_size(self, input_layer_config):
        return self.size
Q
qijun 已提交
529

Z
zhangjinchao01 已提交
530 531
    def calc_parameter_size(self, input_size, output_size):
        raise NotimplementedError
Q
qijun 已提交
532

Z
zhangjinchao01 已提交
533 534 535 536 537 538 539 540 541 542
    def calc_parameter_dims(self, input_size, output_size):
        raise NotimplementedError


@config_class
class IdentityProjection(Projection):
    type = 'identity'

    def calc_output_size(self, input_layer_config):
        return input_layer_config.size
Q
qijun 已提交
543

Z
zhangjinchao01 已提交
544 545
    def calc_parameter_size(self, input_size, output_size):
        return 0
Q
qijun 已提交
546

Z
zhangjinchao01 已提交
547 548 549
    def calc_parameter_dims(self, input_size, output_size):
        return []

Q
qijun 已提交
550

Z
zhangjinchao01 已提交
551 552 553 554 555 556
# Like IdentityProjection, but layer size may smaller than input size,
# the projection select dimesions [offset, offset+layer_size) from input
@config_class
class IdentityOffsetProjection(Projection):
    type = 'identity_offset'

Q
qijun 已提交
557 558 559
    def __init__(self, input_layer_name, offset, **xargs):
        super(IdentityOffsetProjection, self).__init__(input_layer_name,
                                                       **xargs)
Z
zhangjinchao01 已提交
560 561 562 563
        self.proj_conf.offset = offset

    def calc_parameter_size(self, input_size, output_size):
        return 0
Q
qijun 已提交
564

Z
zhangjinchao01 已提交
565 566 567
    def calc_parameter_dims(self, input_size, output_size):
        return []

Q
qijun 已提交
568

569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597
@config_class
class SliceProjection(Projection):
    type = 'slice'

    def __init__(self, input_layer_name, slices, **xargs):
        super(SliceProjection, self).__init__(input_layer_name, **xargs)
        input = g_layer_map[input_layer_name]
        if input.type in ["exconv", "cudnn_conv"]:
            # the slice operator is for the channel dimension
            assert input.num_filters is not None
            channels = input.num_filters
            image_size = input.size / channels
            assert slices[len(slices) - 1][1] <= channels
            for i in xrange(len(slices)):
                slice = self.proj_conf.slices.add()
                slice.start = slices[i][0] * image_size
                slice.end = slices[i][1] * image_size
                self.size += slice.end - slice.start
        else:
            config_assert(False,
                          'Currently the input should be convolution layer')

    def calc_parameter_size(self, input_size, output_size):
        return 0

    def calc_parameter_dims(self, input_size, output_size):
        return []


Z
zhangjinchao01 已提交
598 599 600 601 602 603 604
# DotMulProjection performs element-wise multiplication with weight
@config_class
class DotMulProjection(Projection):
    type = 'dot_mul'

    def calc_output_size(self, input_layer_config):
        return input_layer_config.size
Q
qijun 已提交
605

Z
zhangjinchao01 已提交
606 607
    def calc_parameter_size(self, input_size, output_size):
        return output_size
Q
qijun 已提交
608

Z
zhangjinchao01 已提交
609 610 611
    def calc_parameter_dims(self, input_size, output_size):
        return [1, output_size]

L
Luo Tao 已提交
612

X
xuwei06 已提交
613 614 615 616 617 618 619 620 621 622 623 624 625 626
# ScalingProjection
@config_class
class ScalingProjection(Projection):
    type = 'scaling'

    def calc_output_size(self, input_layer_config):
        return input_layer_config.size

    def calc_parameter_size(self, input_size, output_size):
        return 1

    def calc_parameter_dims(self, input_size, output_size):
        return [1, 1]

Q
qijun 已提交
627

Z
zhangjinchao01 已提交
628 629 630 631 632 633
@config_class
class TableProjection(Projection):
    type = 'table'

    def calc_parameter_size(self, input_size, output_size):
        return input_size * output_size
Q
qijun 已提交
634

Z
zhangjinchao01 已提交
635 636 637
    def calc_parameter_dims(self, input_size, output_size):
        return [input_size, output_size]

Q
qijun 已提交
638

Z
zhangjinchao01 已提交
639 640 641 642 643 644
@config_class
class FullMatrixProjection(Projection):
    type = 'fc'

    def calc_parameter_size(self, input_size, output_size):
        return input_size * output_size
Q
qijun 已提交
645

Z
zhangjinchao01 已提交
646 647 648
    def calc_parameter_dims(self, input_size, output_size):
        return [input_size, output_size]

Q
qijun 已提交
649

Z
zhangjinchao01 已提交
650 651 652 653 654 655
@config_class
class TransposedFullMatrixProjection(Projection):
    type = 'trans_fc'

    def calc_parameter_size(self, input_size, output_size):
        return input_size * output_size
Q
qijun 已提交
656

Z
zhangjinchao01 已提交
657 658 659
    def calc_parameter_dims(self, input_size, output_size):
        return [output_size, input_size]

Q
qijun 已提交
660

Z
zhangjinchao01 已提交
661 662 663 664
@config_class
class ContextProjection(Projection):
    type = 'context'

Q
qijun 已提交
665 666
    def __init__(self, input_layer_name, context_start, context_length,
                 trainable_padding, **xargs):
Z
zhangjinchao01 已提交
667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689
        super(ContextProjection, self).__init__(input_layer_name, **xargs)
        self.proj_conf.context_start = context_start
        self.proj_conf.context_length = context_length
        self.proj_conf.trainable_padding = trainable_padding
        self._total_pad = max(0, -self.proj_conf.context_start) \
                          + max(0, self.proj_conf.context_start \
                                + self.proj_conf.context_length - 1)

    def calc_output_size(self, input_layer_config):
        return input_layer_config.size * self.proj_conf.context_length

    def calc_parameter_size(self, input_size, output_size):
        if self.proj_conf.trainable_padding == False:
            return 0
        else:
            return input_size * self._total_pad

    def calc_parameter_dims(self, input_size, output_size):
        return [self._total_pad, input_size]

    _total_pad = 0


690
@config_class
691
class ConvBaseProjection(Projection):
Q
qijun 已提交
692 693 694 695 696
    def __init__(self,
                 input_layer_name,
                 num_filters=None,
                 conv_conf=None,
                 **xargs):
697
        super(ConvBaseProjection, self).__init__(input_layer_name, **xargs)
698 699 700 701 702 703 704 705 706 707 708 709

        if num_filters is not None:
            self.proj_conf.num_filters = num_filters

    def calc_output_size(self, input_layer_config):
        return self.proj_conf.output_size

    def calc_parameter_size(self, input_size, output_size):
        co = self.proj_conf.num_filters
        ci = self.proj_conf.conv_conf.channels
        fh = self.proj_conf.conv_conf.filter_size
        fw = self.proj_conf.conv_conf.filter_size_y
710 711
        gr = self.proj_conf.conv_conf.groups
        return co * ci * fh * fw / gr
712 713 714 715 716 717 718

    def calc_bias_size(self):
        return self.proj_conf.num_filters

    def calc_parameter_dims(self, input_size, output_size):
        return None

Q
qijun 已提交
719

720 721 722 723 724 725 726 727 728
@config_class
class ConvProjection(ConvBaseProjection):
    type = 'conv'

    def __init__(self,
                 input_layer_name,
                 num_filters=None,
                 conv_conf=None,
                 **xargs):
729 730
        super(ConvProjection, self).__init__(input_layer_name, num_filters,
                                             conv_conf, **xargs)
731

732
        parse_conv(conv_conf, self.input_layer_name, self.proj_conf.conv_conf,
733 734 735 736 737 738 739 740 741 742 743 744 745 746 747
                   num_filters)
        self.proj_conf.output_size = self.proj_conf.conv_conf.output_x * \
                                     self.proj_conf.conv_conf.output_y * \
                                     num_filters


@config_class
class ConvTransProjection(ConvBaseProjection):
    type = 'convt'

    def __init__(self,
                 input_layer_name,
                 num_filters=None,
                 conv_conf=None,
                 **xargs):
748 749
        super(ConvTransProjection, self).__init__(input_layer_name, num_filters,
                                                  conv_conf, **xargs)
750 751 752

        parse_conv(
            conv_conf,
753
            self.input_layer_name,
754 755 756 757 758 759 760 761
            self.proj_conf.conv_conf,
            num_filters,
            trans=True)
        self.proj_conf.output_size = self.proj_conf.conv_conf.img_size_y * \
                                     self.proj_conf.conv_conf.img_size * \
                                     num_filters


Z
zhangjinchao01 已提交
762 763 764
# Define a operator for mixed layer
@config_class
class Operator(Cfg):
Q
qijun 已提交
765 766
    type = None  # subclass should set it correctly

Z
zhangjinchao01 已提交
767 768
    def __init__(
            self,
Q
qijun 已提交
769
            input_layer_names, ):
Z
zhangjinchao01 已提交
770 771 772 773 774 775 776 777 778 779
        self.add_keys(locals())
        self.operator_conf = OperatorConfig()
        self.operator_conf.type = self.type

    def check_dims(self):
        pass

    def calc_output_size(self, input_sizes):
        return 0

Q
qijun 已提交
780

Z
zhangjinchao01 已提交
781 782 783
@config_class
class DotMulOperator(Operator):
    type = 'dot_mul'
Q
qijun 已提交
784 785 786

    def __init__(self, input_layer_names, scale=None, **xargs):
        super(DotMulOperator, self).__init__(input_layer_names, **xargs)
Z
zhangjinchao01 已提交
787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804
        if scale is not None:
            self.operator_conf.dotmul_scale = scale

        config_assert(len(input_layer_names) == 2, "DotMul is binary operator")

    def check_dims(self):
        for i in range(2):
            config_assert(self.operator_conf.input_sizes[i] ==
                          self.operator_conf.output_size,
                          "DotMul input_size != output_size")

    def calc_output_size(self, input_sizes):
        return input_sizes[0]


@config_class
class ConvOperator(Operator):
    type = 'conv'
Q
qijun 已提交
805 806 807 808 809 810 811

    def __init__(self,
                 input_layer_names,
                 num_filters=None,
                 conv_conf=None,
                 **xargs):
        super(ConvOperator, self).__init__(input_layer_names, **xargs)
Z
zhangjinchao01 已提交
812 813 814
        if num_filters is not None:
            self.operator_conf.num_filters = num_filters

815 816
        parse_conv(conv_conf,
                   MakeLayerNameInSubmodel(input_layer_names[0]),
Q
qijun 已提交
817
                   self.operator_conf.conv_conf, num_filters)
L
Luo Tao 已提交
818 819 820
        self.operator_conf.output_size = self.operator_conf.conv_conf.output_x * \
                                         self.operator_conf.conv_conf.output_y * \
                                         num_filters
Z
zhangjinchao01 已提交
821 822 823

        config_assert(len(input_layer_names) == 2, "Conv is binary operator")

824 825
    def calc_output_size(self, input_sizes):
        return self.operator_conf.output_size
Z
zhangjinchao01 已提交
826 827


828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857
@config_class
class ConvTransOperator(Operator):
    type = 'convt'

    def __init__(self,
                 input_layer_names,
                 num_filters=None,
                 conv_conf=None,
                 **xargs):
        super(ConvTransOperator, self).__init__(input_layer_names, **xargs)
        if num_filters is not None:
            self.operator_conf.num_filters = num_filters

        parse_conv(
            conv_conf,
            MakeLayerNameInSubmodel(input_layer_names[0]),
            self.operator_conf.conv_conf,
            num_filters,
            trans=True)
        self.operator_conf.output_size = \
            self.operator_conf.conv_conf.img_size * \
            self.operator_conf.conv_conf.img_size_y * \
            num_filters

        config_assert(len(input_layer_names) == 2, "Conv is binary operator")

    def calc_output_size(self, input_sizes):
        return self.operator_conf.output_size


Z
zhangjinchao01 已提交
858 859 860
# please refer to the comments in proto/ModelConfig.proto
@config_class
class Conv(Cfg):
Q
qijun 已提交
861 862 863 864 865 866 867 868 869 870 871 872
    def __init__(self,
                 filter_size,
                 channels,
                 padding=None,
                 stride=None,
                 groups=None,
                 filter_channels=None,
                 output_x=None,
                 img_size=None,
                 caffe_mode=True,
                 filter_size_y=None,
                 padding_y=None,
W
wanghaoshuang 已提交
873 874 875
                 stride_y=None,
                 dilation=None,
                 dilation_y=None):
Z
zhangjinchao01 已提交
876 877
        self.add_keys(locals())
        if filter_size_y is None:
Q
qijun 已提交
878
            self.filter_size_y = filter_size
Z
zhangjinchao01 已提交
879
        if padding_y is None:
Q
qijun 已提交
880
            self.padding_y = padding
881 882
        if dilation_y is None:
            self.dilation_y = dilation
Z
zhangjinchao01 已提交
883
        if stride_y is None:
Q
qijun 已提交
884
            self.stride_y = stride
Z
zhangjinchao01 已提交
885
        if output_x is not None:
Q
qijun 已提交
886 887
            config_assert(output_x <= 0)

Z
zhangjinchao01 已提交
888

889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908
# please refer to the comments in proto/ModelConfig.proto
@config_class
class Conv3D(Cfg):
    def __init__(self,
                 filter_size,
                 channels,
                 padding=None,
                 stride=None,
                 groups=None,
                 filter_channels=None,
                 output_x=None,
                 img_size=None,
                 caffe_mode=True,
                 filter_size_y=None,
                 padding_y=None,
                 stride_y=None,
                 filter_size_z=None,
                 padding_z=None,
                 stride_z=None):
        self.add_keys(locals())
C
chengduoZH 已提交
909 910 911 912 913 914
        self.filter_size_y = filter_size_y if filter_size_y else filter_size
        self.filter_size_z = filter_size_z if filter_size_z else filter_size
        self.padding_y = padding_y if padding_y else padding
        self.padding_z = padding_z if padding_z else padding
        self.stride_y = stride_y if stride_y else stride
        self.stride_z = stride_z if stride_z else stride
915 916 917 918
        if output_x is not None:
            config_assert(output_x <= 0)


L
liaogang 已提交
919 920
@config_class
class BilinearInterp(Cfg):
L
Luo Tao 已提交
921
    def __init__(self, out_size_x=None, out_size_y=None, channels=None):
L
liaogang 已提交
922 923
        self.add_keys(locals())

Q
qijun 已提交
924

Z
zhangjinchao01 已提交
925 926
@config_class
class Pool(Cfg):
D
dangqingqing 已提交
927 928 929 930 931 932 933 934 935 936 937
    def __init__(
            self,
            pool_type,
            channels,
            size_x,
            size_y=None,
            start=None,
            stride=None,  # 1 by defalut in protobuf
            stride_y=None,
            padding=None,  # 0 by defalut in protobuf
            padding_y=None):
Z
zhangjinchao01 已提交
938
        self.add_keys(locals())
Q
qijun 已提交
939 940


C
chengduoZH 已提交
941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965
@config_class
class Pool3d(Cfg):
    def __init__(
            self,
            pool_type,
            channels,
            size_x,
            size_y=None,
            size_z=None,
            start=None,
            stride=None,  # 1 by defalut in protobuf
            stride_y=None,
            stride_z=None,
            padding=None,  # 0 by defalut in protobuf
            padding_y=None,
            padding_z=None):
        self.add_keys(locals())
        self.filter_size_y = size_y if size_y else size_x
        self.filter_size_z = size_z if size_z else size_x
        self.padding_y = padding_y if padding_y else padding
        self.padding_z = padding_z if padding_z else padding
        self.stride_y = stride_y if stride_y else stride
        self.stride_z = stride_z if stride_z else stride


Q
qijun 已提交
966
@config_class
Q
qijun 已提交
967
class SpatialPyramidPool(Cfg):
L
Luo Tao 已提交
968
    def __init__(self, pool_type, pyramid_height, channels):
Q
qijun 已提交
969
        self.add_keys(locals())
Z
zhangjinchao01 已提交
970

Q
qijun 已提交
971

D
dangqingqing 已提交
972 973 974 975 976 977
@config_class
class Pad(Cfg):
    def __init__(self, channels, pad_c, pad_h, pad_w):
        self.add_keys(locals())


Z
zhangjinchao01 已提交
978 979
@config_class
class Norm(Cfg):
Q
qijun 已提交
980 981 982 983 984 985 986 987 988
    def __init__(self,
                 norm_type,
                 channels,
                 size,
                 scale,
                 pow,
                 output_x=None,
                 img_size=None,
                 blocked=None):
Z
zhangjinchao01 已提交
989 990
        self.add_keys(locals())

Q
qijun 已提交
991

Z
zhangjinchao01 已提交
992 993
@config_class
class Image(Cfg):
Q
qijun 已提交
994
    def __init__(self, channels, img_size=None):
Z
zhangjinchao01 已提交
995 996
        self.add_keys(locals())

Q
qijun 已提交
997

Z
zhangjinchao01 已提交
998 999
@config_class
class BlockExpand(Cfg):
Q
qijun 已提交
1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011
    def __init__(self,
                 channels,
                 padding_x=0,
                 padding_y=0,
                 stride_x=0,
                 stride_y=0,
                 block_x=0,
                 block_y=0,
                 img_size_x=0,
                 img_size_y=0,
                 output_x=0,
                 output_y=0):
Z
zhangjinchao01 已提交
1012 1013
        self.add_keys(locals())

Q
qijun 已提交
1014

1015 1016
@config_class
class MaxOut(Cfg):
Q
qijun 已提交
1017
    def __init__(self, channels, groups, img_size_x=0, img_size_y=0):
1018 1019
        self.add_keys(locals())

Q
qijun 已提交
1020

1021
def create_data_config_proto(async_load_data=False,
1022
                             constant_slots=None,
王益 已提交
1023 1024 1025
                             data_ratio=1,
                             is_main_data=True,
                             usage_ratio=None):
Z
zhangjinchao01 已提交
1026 1027 1028 1029 1030 1031 1032 1033
    # default: all sub dataproviders are treat as "main data".
    # see proto/DataConfig.proto for is_main_data
    data_config = DataConfig()

    data_config.async_load_data = async_load_data

    if constant_slots:
        data_config.constant_slots.extend(constant_slots)
Q
qijun 已提交
1034 1035
    data_config.data_ratio = data_ratio
    data_config.is_main_data = is_main_data
Z
zhangjinchao01 已提交
1036

Q
qijun 已提交
1037
    usage_ratio = default(usage_ratio, settings_deprecated["usage_ratio"])
Z
zhangjinchao01 已提交
1038 1039 1040 1041 1042 1043
    config_assert(usage_ratio >= 0 and usage_ratio <= 1,
                  "The range of usage_ratio is [0, 1]")
    data_config.usage_ratio = usage_ratio

    return data_config

Q
qijun 已提交
1044

Z
zhangjinchao01 已提交
1045
@config_func
Q
qijun 已提交
1046 1047 1048 1049 1050
def SimpleData(files=None,
               feat_dim=None,
               context_len=None,
               buffer_capacity=None,
               **xargs):
1051
    data_config = create_data_config_proto(**xargs)
Z
zhangjinchao01 已提交
1052 1053 1054 1055 1056 1057 1058 1059 1060
    data_config.type = 'simple'
    data_config.files = files
    data_config.feat_dim = feat_dim
    if context_len is not None:
        data_config.context_len = context_len
    if buffer_capacity:
        data_config.buffer_capacity = buffer_capacity
    return data_config

Q
qijun 已提交
1061

Z
zhangjinchao01 已提交
1062
@config_func
Q
qijun 已提交
1063 1064 1065 1066 1067 1068 1069 1070 1071 1072
def PyData(files=None,
           type=None,
           file_group_queue_capacity=None,
           load_data_module=None,
           load_data_object=None,
           load_data_args="",
           load_file_count=None,
           constant_slots=None,
           load_thread_num=None,
           **xargs):
1073
    data_config = create_data_config_proto(**xargs)
Z
zhangjinchao01 已提交
1074 1075
    data_config.type = 'py'
    if load_data_module in g_py_module_name_list:
Q
qijun 已提交
1076

Z
zhangjinchao01 已提交
1077 1078 1079
        def get_path(module):
            m = __import__(load_data_module)
            return os.path.split(os.path.realpath(m.__file__))[0]
Q
qijun 已提交
1080

Z
zhangjinchao01 已提交
1081 1082 1083
        # python C-api is not thread safe, one module can only be import once,
        # so here we nedd to copy the module with different names if it has to be
        # imported several times.
Q
qijun 已提交
1084 1085
        module_new_name = "%s_copy_%d" % (load_data_module,
                                          len(g_py_module_name_list))
Z
zhangjinchao01 已提交
1086
        g_py_module_name_list.append(module_new_name)
Q
qijun 已提交
1087 1088 1089 1090
        module_path = "%s/%s.py" % (get_path(load_data_module),
                                    load_data_module)
        new_module_path = "%s/%s.py" % (get_path(load_data_module),
                                        module_new_name)
Z
zhangjinchao01 已提交
1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114
        if os.path.isfile(module_path) == False:
            raise Exception("File %s is not exist." % module_path)
        shutil.copy2(module_path, new_module_path)
        load_data_module = module_new_name
    else:
        g_py_module_name_list.append(load_data_module)
    if load_data_module is not None and load_data_object is not None:
        data_config.load_data_module = load_data_module
        data_config.load_data_object = load_data_object
    else:
        raise ValueError('load_data_module, load_data_object is not defined.')
    data_config.load_data_args = load_data_args

    data_config.files = files or ''
    if file_group_queue_capacity is not None:
        data_config.file_group_conf.queue_capacity = file_group_queue_capacity
    if load_file_count is not None:
        data_config.file_group_conf.load_file_count = load_file_count
    if load_thread_num is not None:
        data_config.file_group_conf.load_thread_num = load_thread_num
    if constant_slots:
        data_config.constant_slots.extend(constant_slots)
    return data_config

Q
qijun 已提交
1115

Z
zhangjinchao01 已提交
1116
@config_func
Q
qijun 已提交
1117 1118 1119 1120 1121 1122 1123
def ProtoData(files=None,
              type=None,
              file_group_queue_capacity=None,
              load_file_count=None,
              constant_slots=None,
              load_thread_num=None,
              **xargs):
1124
    data_config = create_data_config_proto(**xargs)
Z
zhangjinchao01 已提交
1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143
    if type is None:
        data_config.type = 'proto'
    else:
        data_config.type = type
    data_config.files = files

    # When type="proto_group", one data provider contains at most
    # load_file_count files, and there are at most
    # (queue_capacity + load_thread_num + 1) data providers in memory
    if file_group_queue_capacity is not None:
        data_config.file_group_conf.queue_capacity = file_group_queue_capacity
    if load_file_count is not None:
        data_config.file_group_conf.load_file_count = load_file_count
    if load_thread_num is not None:
        data_config.file_group_conf.load_thread_num = load_thread_num
    if constant_slots:
        data_config.constant_slots.extend(constant_slots)
    return data_config

Q
qijun 已提交
1144

Z
zhangjinchao01 已提交
1145 1146
#real data for training is actually provided by "sub_data" data providers.
@config_func
Q
qijun 已提交
1147
def MultiData(sub_data=[]):
Z
zhangjinchao01 已提交
1148 1149 1150 1151 1152
    data_config = DataConfig()
    data_config.type = 'multi'
    data_config.sub_data_configs.extend(sub_data)
    return data_config

Q
qijun 已提交
1153

Z
zhangjinchao01 已提交
1154
@config_func
Q
qijun 已提交
1155 1156 1157 1158 1159 1160 1161
def Data(type,
         files=None,
         feat_dim=None,
         slot_dims=None,
         context_len=None,
         buffer_capacity=None,
         **xargs):
Z
zhangjinchao01 已提交
1162

1163
    data_config = create_data_config_proto(**xargs)
Z
zhangjinchao01 已提交
1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196
    data_config.type = type
    data_config.files = files
    data_config.feat_dim = feat_dim
    data_config.slot_dims.extend(slot_dims)
    if context_len is not None:
        data_config.context_len = context_len
    data_config.buffer_capacity = buffer_capacity
    return data_config


@config_func
def TrainData(data_config, async_load_data=None):
    config_assert(not g_config.HasField('data_config'),
                  'Only one TrainData definition is allowed')
    g_config.data_config.CopyFrom(data_config)
    g_config.data_config.for_test = False
    if async_load_data is not None:
        logger.warning("Deprecated: async_load_data should be used inside"
                       " Data definition")
        g_config.data_config.async_load_data = async_load_data


@config_func
def TestData(data_config, async_load_data=None):
    config_assert(not g_config.HasField('test_data_config'),
                  'Only one TestData definition is allowed')
    g_config.test_data_config.CopyFrom(data_config)
    g_config.test_data_config.for_test = True
    if async_load_data is not None:
        logger.warning("Deprecated: async_load_data should be used inside"
                       " Data definition")
        g_config.test_data_config.async_load_data = async_load_data

Q
qijun 已提交
1197

L
Luo Tao 已提交
1198 1199
#caffe_mode: compute the output size using floor instead of ceil,
#            which is consistent of caffe and CuDNN's convention.
1200 1201 1202 1203 1204 1205 1206
def cnn_output_size(img_size, filter_size, padding, stride, caffe_mode):
    output = (2 * padding + img_size - filter_size) / float(stride)
    if caffe_mode:
        return 1 + int(math.floor(output))
    else:
        return 1 + int(math.ceil(output))

Q
qijun 已提交
1207

1208
#calcualte image_size based on output_size for de-convolution (ConvTransLayer).
L
Luo Tao 已提交
1209
#It is the reverse function of cnn_output_size
1210
def cnn_image_size(output_size, filter_size, padding, stride, caffe_mode):
L
Luo Tao 已提交
1211 1212 1213
    img_size = (output_size - 1) * stride + filter_size - 2 * padding
    if not caffe_mode:
        img_size = img_size + 1
1214 1215
    return img_size

Q
qijun 已提交
1216

L
Luo Tao 已提交
1217
def get_img_size(input_layer_name, channels):
L
Luo Tao 已提交
1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229
    input = g_layer_map[input_layer_name]
    img_pixels = input.size / channels
    img_size = input.width if input.width > 0 else int(img_pixels**0.5)
    img_size_y = input.height if input.height > 0 else int(img_pixels /
                                                           img_size)
    config_assert(
        img_size * img_size_y == img_pixels,
        "Input layer %s: Incorrect input image size %d * %d for input image pixels %d"
        % (input_layer_name, img_size, img_size_y, img_pixels))
    return img_size, img_size_y


C
chengduoZH 已提交
1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243
def get_img3d_size(input_layer_name, channels):
    input = g_layer_map[input_layer_name]
    img_pixels = input.size / channels
    img_size = input.width
    img_size_y = input.height
    img_size_z = input.depth

    config_assert(
        img_size * img_size_y * img_size_z == img_pixels,
        "Input layer %s: Incorrect input image size %d * %d * %d for input image pixels %d"
        % (input_layer_name, img_size, img_size_y, img_size_z, img_pixels))
    return img_size, img_size_y, img_size_z


L
Luo Tao 已提交
1244 1245 1246 1247 1248 1249
def parse_bilinear(bilinear, input_layer_name, bilinear_conf):
    parse_image(bilinear, input_layer_name, bilinear_conf.image_conf)
    bilinear_conf.out_size_x = bilinear.out_size_x
    bilinear_conf.out_size_y = bilinear.out_size_y


1250
def parse_pool(pool, input_layer_name, pool_conf, ceil_mode):
Z
zhangjinchao01 已提交
1251
    pool_conf.pool_type = pool.pool_type
Q
qijun 已提交
1252 1253 1254
    config_assert(pool.pool_type in [
        'max-projection', 'avg-projection', 'cudnn-max-pool', 'cudnn-avg-pool'
    ], "pool-type %s is not in "
Z
zhangjinchao01 已提交
1255
                  "['max-projection', 'avg-projection', "
Q
qijun 已提交
1256
                  "'cudnn-max-pool', 'cudnn-avg-pool']" % pool.pool_type)
Z
zhangjinchao01 已提交
1257 1258 1259 1260 1261 1262

    pool_conf.channels = pool.channels
    pool_conf.size_x = pool.size_x
    pool_conf.stride = pool.stride

    pool_conf.size_y = default(pool.size_y, pool_conf.size_x)
Q
qijun 已提交
1263
    pool_conf.stride_y = default(pool.stride_y, pool_conf.stride)
Z
zhangjinchao01 已提交
1264

L
Luo Tao 已提交
1265
    pool_conf.img_size, pool_conf.img_size_y = \
L
Luo Tao 已提交
1266
        get_img_size(input_layer_name, pool.channels)
Z
zhangjinchao01 已提交
1267

1268
    config_assert(not pool.start, "start is deprecated in pooling.")
Z
zhangjinchao01 已提交
1269

1270
    if pool.padding is not None:
Z
zhangjinchao01 已提交
1271
        pool_conf.padding = pool.padding
1272
    pool_conf.padding_y = default(pool.padding_y, pool_conf.padding)
D
dangqingqing 已提交
1273 1274
    pool_conf.output_x = cnn_output_size(pool_conf.img_size, pool_conf.size_x,
                                         pool_conf.padding, pool_conf.stride,
1275
                                         not ceil_mode)
D
dangqingqing 已提交
1276 1277
    pool_conf.output_y = cnn_output_size(pool_conf.img_size_y, pool_conf.size_y,
                                         pool_conf.padding_y,
1278
                                         pool_conf.stride_y, not ceil_mode)
Q
qijun 已提交
1279

Z
zhangjinchao01 已提交
1280

C
chengduoZH 已提交
1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319
def parse_pool3d(pool, input_layer_name, pool_conf, ceil_mode):
    pool_conf.pool_type = pool.pool_type
    config_assert(pool.pool_type in ['max-projection', 'avg-projection'],
                  "pool-type %s is not in "
                  "['max-projection', 'avg-projection']" % pool.pool_type)

    pool_conf.channels = pool.channels

    pool_conf.size_x = pool.size_x
    pool_conf.stride = pool.stride
    pool_conf.padding = pool.padding

    pool_conf.size_y = default(pool.size_y, pool_conf.size_x)
    pool_conf.size_z = default(pool.size_z, pool_conf.size_x)
    pool_conf.stride_y = default(pool.stride_y, pool_conf.stride)
    pool_conf.stride_z = default(pool.stride_z, pool_conf.stride)
    pool_conf.padding_y = default(pool.padding_y, pool_conf.padding)
    pool_conf.padding_z = default(pool.padding_z, pool_conf.padding)

    pool_conf.img_size, pool_conf.img_size_y, pool_conf.img_size_z = \
        get_img3d_size(input_layer_name, pool.channels)

    config_assert(not pool.start, "start is deprecated in pooling.")

    if pool.padding is not None:
        pool_conf.padding = pool.padding
    pool_conf.padding_y = default(pool.padding_y, pool_conf.padding)
    pool_conf.padding_z = default(pool.padding_z, pool_conf.padding)
    pool_conf.output_x = cnn_output_size(pool_conf.img_size, pool_conf.size_x,
                                         pool_conf.padding, pool_conf.stride,
                                         not ceil_mode)
    pool_conf.output_y = cnn_output_size(pool_conf.img_size_y, pool_conf.size_y,
                                         pool_conf.padding_y,
                                         pool_conf.stride_y, not ceil_mode)
    pool_conf.output_z = cnn_output_size(pool_conf.img_size_z, pool_conf.size_z,
                                         pool_conf.padding_z,
                                         pool_conf.stride_z, not ceil_mode)


Q
qijun 已提交
1320
def parse_spp(spp, input_layer_name, spp_conf):
L
Luo Tao 已提交
1321
    parse_image(spp, input_layer_name, spp_conf.image_conf)
Q
qijun 已提交
1322 1323
    spp_conf.pool_type = spp.pool_type
    config_assert(spp.pool_type in ['max-projection', 'avg-projection'],
Q
qijun 已提交
1324 1325
                  "pool-type %s is not in "
                  "['max-projection', 'avg-projection']" % spp.pool_type)
Q
qijun 已提交
1326
    spp_conf.pyramid_height = spp.pyramid_height
Q
qijun 已提交
1327

Q
qijun 已提交
1328

Z
zhangjinchao01 已提交
1329 1330
def parse_image(image, input_layer_name, image_conf):
    image_conf.channels = image.channels
L
Luo Tao 已提交
1331
    image_conf.img_size, image_conf.img_size_y = \
L
Luo Tao 已提交
1332
        get_img_size(input_layer_name, image_conf.channels)
Q
qijun 已提交
1333

Z
zhangjinchao01 已提交
1334

C
chengduoZH 已提交
1335 1336 1337 1338 1339 1340
def parse_image3d(image, input_layer_name, image_conf):
    image_conf.channels = image.channels
    image_conf.img_size, image_conf.img_size_y, image_conf.img_size_z = \
        get_img3d_size(input_layer_name, image_conf.channels)


Z
zhangjinchao01 已提交
1341 1342
def parse_norm(norm, input_layer_name, norm_conf):
    norm_conf.norm_type = norm.norm_type
1343 1344 1345 1346 1347
    config_assert(
        norm.norm_type in
        ['rnorm', 'cmrnorm-projection', 'cross-channel-norm'],
        "norm-type %s is not in [rnorm, cmrnorm-projection, cross-channel-norm]"
        % norm.norm_type)
Z
zhangjinchao01 已提交
1348 1349 1350 1351 1352 1353
    norm_conf.channels = norm.channels
    norm_conf.size = norm.size
    norm_conf.scale = norm.scale
    norm_conf.pow = norm.pow
    norm_conf.blocked = norm.blocked

L
Luo Tao 已提交
1354
    norm_conf.img_size, norm_conf.img_size_y = \
L
Luo Tao 已提交
1355
        get_img_size(input_layer_name, norm.channels)
Z
zhangjinchao01 已提交
1356
    norm_conf.output_x = norm_conf.img_size
L
Luo Tao 已提交
1357
    norm_conf.output_y = norm_conf.img_size_y
Z
zhangjinchao01 已提交
1358 1359 1360
    if norm.norm_type in ['cmrnorm-projection']:
        norm_conf.scale /= norm.size
    else:
Q
qijun 已提交
1361 1362
        norm_conf.scale /= norm.size**2

1363

L
Luo Tao 已提交
1364 1365
#caffe_mode: compute the output size using floor instead of ceil,
#            which is consistent of caffe and CuDNN's convention.
1366
def parse_conv(conv, input_layer_name, conv_conf, num_filters, trans=False):
Z
zhangjinchao01 已提交
1367 1368 1369 1370 1371 1372 1373 1374 1375
    conv_conf.filter_size = conv.filter_size
    conv_conf.filter_size_y = conv.filter_size_y
    conv_conf.channels = conv.channels
    conv_conf.padding = conv.padding
    conv_conf.padding_y = conv.padding_y
    conv_conf.stride = conv.stride
    conv_conf.stride_y = conv.stride_y
    conv_conf.groups = conv.groups
    conv_conf.caffe_mode = conv.caffe_mode
Q
qijun 已提交
1376

1377
    if not trans:
1378
        conv_conf.filter_channels = conv.channels / conv.groups
L
Luo Tao 已提交
1379
        conv_conf.img_size, conv_conf.img_size_y = \
L
Luo Tao 已提交
1380
            get_img_size(input_layer_name, conv.channels)
1381
        conv_conf.output_x = cnn_output_size(
Q
qijun 已提交
1382 1383
            conv_conf.img_size, conv_conf.filter_size, conv_conf.padding,
            conv_conf.stride, conv_conf.caffe_mode)
L
Luo Tao 已提交
1384 1385 1386
        conv_conf.output_y = cnn_output_size(
            conv_conf.img_size_y, conv_conf.filter_size_y, conv_conf.padding_y,
            conv_conf.stride_y, conv_conf.caffe_mode)
1387
    else:
1388
        conv_conf.filter_channels = num_filters / conv.groups
L
Luo Tao 已提交
1389
        conv_conf.output_x, conv_conf.output_y = \
L
Luo Tao 已提交
1390
            get_img_size(input_layer_name, conv.channels)
1391
        conv_conf.img_size = cnn_image_size(
Q
qijun 已提交
1392 1393
            conv_conf.output_x, conv_conf.filter_size, conv_conf.padding,
            conv_conf.stride, conv_conf.caffe_mode)
L
Luo Tao 已提交
1394
        conv_conf.img_size_y = cnn_image_size(
L
Luo Tao 已提交
1395 1396
            conv_conf.output_y, conv_conf.filter_size_y, conv_conf.padding_y,
            conv_conf.stride_y, conv_conf.caffe_mode)
Q
qijun 已提交
1397

1398

1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442
#caffe_mode: compute the output size using floor instead of ceil,
#            which is consistent of caffe and CuDNN's convention.
def parse_conv3d(conv, input_layer_name, conv_conf, num_filters, trans=False):
    conv_conf.filter_size = conv.filter_size
    conv_conf.filter_size_y = conv.filter_size_y
    conv_conf.filter_size_z = conv.filter_size_z
    conv_conf.channels = conv.channels
    conv_conf.padding = conv.padding
    conv_conf.padding_y = conv.padding_y
    conv_conf.padding_z = conv.padding_z
    conv_conf.stride = conv.stride
    conv_conf.stride_y = conv.stride_y
    conv_conf.stride_z = conv.stride_z
    conv_conf.groups = conv.groups
    conv_conf.caffe_mode = conv.caffe_mode

    if not trans:
        conv_conf.filter_channels = conv.channels / conv.groups
        conv_conf.img_size, conv_conf.img_size_y, conv_conf.img_size_z = \
            get_img3d_size(input_layer_name, conv.channels)
        conv_conf.output_x = cnn_output_size(
            conv_conf.img_size, conv_conf.filter_size, conv_conf.padding,
            conv_conf.stride, conv_conf.caffe_mode)
        conv_conf.output_y = cnn_output_size(
            conv_conf.img_size_y, conv_conf.filter_size_y, conv_conf.padding_y,
            conv_conf.stride_y, conv_conf.caffe_mode)
        conv_conf.output_z = cnn_output_size(
            conv_conf.img_size_z, conv_conf.filter_size_z, conv_conf.padding_z,
            conv_conf.stride_z, conv_conf.caffe_mode)
    else:
        conv_conf.filter_channels = num_filters / conv.groups
        conv_conf.output_x, conv_conf.output_y, conv_conf.output_z = \
            get_img3d_size(input_layer_name, conv.channels)
        conv_conf.img_size = cnn_image_size(
            conv_conf.output_x, conv_conf.filter_size, conv_conf.padding,
            conv_conf.stride, conv_conf.caffe_mode)
        conv_conf.img_size_y = cnn_image_size(
            conv_conf.output_y, conv_conf.filter_size_y, conv_conf.padding_y,
            conv_conf.stride_y, conv_conf.caffe_mode)
        conv_conf.img_size_z = cnn_image_size(
            conv_conf.output_z, conv_conf.filter_size_z, conv_conf.padding_z,
            conv_conf.stride_z, conv_conf.caffe_mode)


Z
zhangjinchao01 已提交
1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455
def parse_block_expand(block_expand, input_layer_name, block_expand_conf):
    block_expand_conf.channels = block_expand.channels
    block_expand_conf.stride_x = block_expand.stride_x
    block_expand_conf.stride_y = block_expand.stride_y
    block_expand_conf.padding_x = block_expand.padding_x
    block_expand_conf.padding_y = block_expand.padding_y
    block_expand_conf.block_x = block_expand.block_x
    block_expand_conf.block_y = block_expand.block_y
    block_expand_conf.img_size_x = block_expand.img_size_x
    block_expand_conf.img_size_y = block_expand.img_size_y
    if block_expand_conf.img_size_x == 0:
        block_expand_conf.output_x = 0
    else:
1456
        block_expand_conf.output_x = cnn_output_size(
1457
            block_expand.img_size_x, block_expand.block_x,
1458
            block_expand.padding_x, block_expand.stride_x, False)
Z
zhangjinchao01 已提交
1459 1460

    if block_expand_conf.img_size_y == 0:
1461
        block_expand_conf.output_y = 0
Z
zhangjinchao01 已提交
1462
    else:
1463
        block_expand_conf.output_y = cnn_output_size(
1464
            block_expand.img_size_y, block_expand.block_y,
1465
            block_expand.padding_y, block_expand.stride_y, False)
Z
zhangjinchao01 已提交
1466

Q
qijun 已提交
1467

1468
def parse_maxout(maxout, input_layer_name, maxout_conf):
L
Luo Tao 已提交
1469
    parse_image(maxout, input_layer_name, maxout_conf.image_conf)
1470
    maxout_conf.groups = maxout.groups
1471

Q
qijun 已提交
1472

Z
zhangjinchao01 已提交
1473 1474
# Define an evaluator
@config_func
Y
yangyaming 已提交
1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491
def Evaluator(name,
              type,
              inputs,
              chunk_scheme=None,
              num_chunk_types=None,
              classification_threshold=None,
              positive_label=None,
              dict_file=None,
              result_file=None,
              num_results=None,
              top_k=None,
              delimited=None,
              excluded_chunk_types=None,
              overlap_threshold=None,
              background_id=None,
              evaluate_difficult=None,
              ap_type=None):
Z
zhangjinchao01 已提交
1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505
    evaluator = g_config.model_config.evaluators.add()
    evaluator.type = type
    evaluator.name = MakeLayerNameInSubmodel(name)
    if type_of(inputs) == str:
        inputs = [inputs]

    evaluator.input_layers.extend(
        [MakeLayerNameInSubmodel(name) for name in inputs])

    if chunk_scheme is not None:
        evaluator.chunk_scheme = chunk_scheme
        evaluator.num_chunk_types = num_chunk_types
    g_current_submodel.evaluator_names.append(evaluator.name)

1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516
    if classification_threshold is not None:
        evaluator.classification_threshold = classification_threshold
    if positive_label is not None:
        evaluator.positive_label = positive_label
    if dict_file is not None:
        evaluator.dict_file = dict_file

    if result_file is not None:
        evaluator.result_file = result_file
    if num_results is not None:
        evaluator.num_results = num_results
L
Liang Zhao 已提交
1517 1518
    if top_k is not None:
        evaluator.top_k = top_k
1519 1520
    if delimited is not None:
        evaluator.delimited = delimited
Z
zhangjinchao01 已提交
1521

1522 1523 1524
    if excluded_chunk_types:
        evaluator.excluded_chunk_types.extend(excluded_chunk_types)

Y
yangyaming 已提交
1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536
    if overlap_threshold is not None:
        evaluator.overlap_threshold = overlap_threshold

    if background_id is not None:
        evaluator.background_id = background_id

    if evaluate_difficult is not None:
        evaluator.evaluate_difficult = evaluate_difficult

    if ap_type is not None:
        evaluator.ap_type = ap_type

Q
qijun 已提交
1537

Z
zhangjinchao01 已提交
1538 1539 1540 1541 1542
class LayerBase(object):
    def __init__(
            self,
            name,
            type,
Q
qijun 已提交
1543
            size,  # size can be 0. In this case, subclass should set it.
Z
zhangjinchao01 已提交
1544 1545 1546 1547
            inputs,
            device=None,
            active_type="",
            drop_rate=0.,
C
caoying03 已提交
1548 1549
            coeff=None,
            error_clipping_threshold=None):
Z
zhangjinchao01 已提交
1550
        config_assert('@' not in name,
Q
qijun 已提交
1551
                      "layer name: %s contain special character @" % name)
Z
zhangjinchao01 已提交
1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566
        global g_current_submodel
        name = MakeLayerNameInSubmodel(name)

        config_assert(name not in g_layer_map,
                      'Duplicated layer name: %s' % name)

        self.inputs = copy.deepcopy(inputs)
        self.operators = []

        if self.inputs is None:
            self.inputs = []
        elif type_of(self.inputs) != list:
            self.inputs = [self.inputs]

        self.config = g_config.model_config.layers.add()
1567
        assert isinstance(self.config, LayerConfig)
1568
        use_mkldnn = bool(int(g_command_config_args.get("use_mkldnn", 0)))
T
tensor-tang 已提交
1569
        mkldnn_acts = ['relu', 'tanh', 'softmax']
1570 1571
        if use_mkldnn and active_type in mkldnn_acts:
            active_type = "mkldnn_" + active_type
Z
zhangjinchao01 已提交
1572 1573 1574
        self.config.name = name
        self.config.type = type
        self.config.active_type = active_type
1575 1576
        if coeff is not None:
            self.config.coeff = float(coeff)
Z
zhangjinchao01 已提交
1577 1578 1579 1580 1581 1582 1583
        if size != 0:
            self.config.size = size
        if drop_rate != 0:
            self.config.drop_rate = drop_rate

        if device is not None:
            self.config.device = device
1584
        elif g_default_device is not None:
Z
zhangjinchao01 已提交
1585 1586
            self.config.device = g_default_device

C
caoying03 已提交
1587 1588 1589
        if error_clipping_threshold is not None:
            self.config.error_clipping_threshold = error_clipping_threshold

Z
zhangjinchao01 已提交
1590 1591 1592 1593 1594 1595 1596
        for input_index in xrange(len(self.inputs)):
            input = self.inputs[input_index]
            input_config = None
            input_layer_name = ''
            if type_of(input) == str:
                input_layer_name = input
                input_config = Input(
Q
qijun 已提交
1597 1598
                    input_layer_name=input,
                    parameter_name=gen_parameter_name(name, input_index))
Z
zhangjinchao01 已提交
1599 1600 1601 1602 1603 1604 1605 1606
                input_layer_name = input_config.input_layer_name
            elif isinstance(input, Input):
                input_layer_name = input.input_layer_name
                input_config = input
                if input_config.parameter_name is None:
                    input_config.parameter_name = \
                        gen_parameter_name(name, input_index)
            elif isinstance(input, Operator):
Q
qijun 已提交
1607
                self.operators.append(input)
Z
zhangjinchao01 已提交
1608 1609 1610 1611
                input.operator_conf.input_indices.append(input_index)
                input_config = Input(input.input_layer_names[0])
                input_layer_name = input_config.input_layer_name
            else:
Q
qijun 已提交
1612
                raise ValueError('Wrong type for inputs: %s' % type_of(input))
Z
zhangjinchao01 已提交
1613
            config_assert(input_layer_name in g_layer_map,
Q
qijun 已提交
1614 1615
                          "Unknown input layer '%s' for layer %s" %
                          (input_layer_name, name))
Z
zhangjinchao01 已提交
1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632
            self.inputs[input_index] = input_config
            layer_input = self.config.inputs.add()
            layer_input.input_layer_name = input_config.input_layer_name
            if input_config.input_layer_argument is not None:
                layer_input.input_layer_argument = \
                    input_config.input_layer_argument

        g_layer_map[name] = self.config

        g_current_submodel.layer_names.append(self.config.name)

    def get_input_layer(self, input_index):
        return g_layer_map[self.config.inputs[input_index].input_layer_name]

    # will return the bias created if not *for_self*
    def create_bias_parameter(
            self,
Q
qijun 已提交
1633
            bias,  # True/False or BiasCfg
Z
zhangjinchao01 已提交
1634
            size,
Q
qijun 已提交
1635 1636 1637
            dims=None,
            for_self=True,  # whether create bias for layer self
    ):
Z
zhangjinchao01 已提交
1638 1639 1640 1641 1642 1643

        if size == 0:
            return
        if dims is None:
            dims = [1, size]

Q
qijun 已提交
1644 1645 1646
        config_assert(
            type_of(bias) == bool or type_of(bias) == Bias,
            'Incorrect type for bias: %s' % type_of(bias))
Z
zhangjinchao01 已提交
1647 1648 1649 1650 1651 1652 1653 1654 1655

        if type_of(bias) == bool:
            if bias:
                bias = Bias()

        if type_of(bias) == Bias:
            if bias.parameter_name is None:
                bias.parameter_name = gen_bias_parameter_name(self.config.name)
            if bias.parameter_name not in g_parameter_map:
1656 1657
                assert isinstance(self.config, LayerConfig)

Z
zhangjinchao01 已提交
1658 1659 1660
                Parameter(
                    bias.parameter_name,
                    size,
Q
qijun 已提交
1661 1662
                    self.config.device
                    if self.config.HasField('device') else None,
Z
zhangjinchao01 已提交
1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673
                    dims,
                    bias.learning_rate,
                    bias.momentum,
                    decay_rate=bias.decay_rate,
                    decay_rate_l1=bias.decay_rate_l1,
                    initial_mean=bias.initial_mean,
                    initial_std=bias.initial_std,
                    initial_strategy=bias.initial_strategy,
                    initial_smart=bias.initial_smart,
                    num_batches_regularization=bias.num_batches_regularization,
                    sparse_remote_update=bias.sparse_remote_update,
Q
qijun 已提交
1674 1675
                    gradient_clipping_threshold=bias.
                    gradient_clipping_threshold,
Z
zhangjinchao01 已提交
1676
                    is_static=bias.is_static,
X
xuwei06 已提交
1677 1678
                    is_shared=bias.is_shared,
                    initializer=bias.initializer)
Z
zhangjinchao01 已提交
1679 1680 1681 1682 1683
            if for_self:
                self.config.bias_parameter_name = bias.parameter_name
            else:
                return bias.parameter_name

Q
qijun 已提交
1684 1685 1686 1687 1688 1689
    def create_input_parameter(self,
                               input_index,
                               size,
                               dims=None,
                               sparse=None,
                               format=None):
Z
zhangjinchao01 已提交
1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703
        if dims is None:
            # TODO(yuyang18): print warning and callstack here!
            dims = list()

        if size == 0:
            return

        input_config = self.inputs[input_index]

        self.config.inputs[input_index].input_parameter_name = \
            input_config.parameter_name

        if input_config.parameter_name in g_parameter_map:
            para = g_parameter_map[input_config.parameter_name]
Q
qijun 已提交
1704 1705
            config_assert(size == para.size, (
                'Shared parameter "%s" does not ' + 'have same size: %s vs. %s')
Z
zhangjinchao01 已提交
1706 1707
                          % (input_config.parameter_name, para.size, size))

Q
qijun 已提交
1708 1709
            config_assert(dims == para.dims, (
                'Shared parameter "%s" does not ' + 'have same dims: %s vs. %s')
Z
zhangjinchao01 已提交
1710 1711 1712 1713 1714 1715
                          % (input_config.parameter_name, para.dims, dims))
            return

        Parameter(
            input_config.parameter_name,
            size,
1716
            self.config.device if self.config.HasField("device") else None,
Z
zhangjinchao01 已提交
1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728
            dims,
            input_config.learning_rate,
            input_config.momentum,
            decay_rate=input_config.decay_rate,
            decay_rate_l1=input_config.decay_rate_l1,
            initial_mean=input_config.initial_mean,
            initial_std=input_config.initial_std,
            initial_strategy=input_config.initial_strategy,
            initial_smart=input_config.initial_smart,
            num_batches_regularization=input_config.num_batches_regularization,
            sparse_remote_update=input_config.sparse_remote_update,
            sparse_update=input_config.sparse_update,
Q
qijun 已提交
1729 1730
            gradient_clipping_threshold=input_config.
            gradient_clipping_threshold,
Z
zhangjinchao01 已提交
1731 1732 1733 1734
            sparse=sparse,
            format=format,
            is_static=input_config.is_static,
            is_shared=input_config.is_shared,
X
xuwei06 已提交
1735 1736
            update_hooks=input_config.update_hooks,
            initializer=input_config.initializer)
Z
zhangjinchao01 已提交
1737 1738 1739 1740 1741 1742 1743 1744 1745

    def set_layer_size(self, size):
        if self.config.size == 0:
            self.config.size = size
        else:
            config_assert(self.config.size == size,
                          'Different inputs result in' +
                          'different layer size at layer %s' % self.config.name)

L
Luo Tao 已提交
1746 1747 1748 1749
    def set_layer_height_width(self, height, width):
        self.config.height = height
        self.config.width = width

C
chengduoZH 已提交
1750 1751 1752
    def set_layer_depth(self, depth):
        self.config.depth = depth

L
Luo Tao 已提交
1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765
    def set_cnn_layer(self,
                      input_layer_name,
                      height,
                      width,
                      channels,
                      is_print=True):
        size = height * width * channels
        self.set_layer_size(size)
        self.set_layer_height_width(height, width)
        if is_print:
            print("output for %s: c = %d, h = %d, w = %d, size = %d" %
                  (input_layer_name, channels, height, width, size))

Q
qijun 已提交
1766

Z
zhangjinchao01 已提交
1767 1768
@config_layer('multi_class_cross_entropy_with_selfnorm')
class MultiClassCrossEntropySelfNormCostLayer(LayerBase):
Q
qijun 已提交
1769 1770 1771
    def __init__(self, name, inputs, softmax_selfnorm_alpha=0.1, **xargs):
        super(MultiClassCrossEntropySelfNormCostLayer, self).__init__(
            name, 'multi_class_cross_entropy_with_selfnorm', 0, inputs, **xargs)
Z
zhangjinchao01 已提交
1772 1773
        self.config.softmax_selfnorm_alpha = softmax_selfnorm_alpha

Q
qijun 已提交
1774

C
caoying03 已提交
1775 1776 1777
@config_layer('cross_entropy_over_beam')
class CrossEntropyOverBeamLayer(LayerBase):
    def __init__(self, name, inputs, **xargs):
C
caoying03 已提交
1778
        config_assert(len(inputs) % 3 == 0, "Error input number.")
C
caoying03 已提交
1779 1780 1781 1782
        super(CrossEntropyOverBeamLayer, self).__init__(
            name, 'cross_entropy_over_beam', 0, inputs, **xargs)
        input_num = len(inputs) / 3
        for i in range(input_num):
C
caoying03 已提交
1783 1784 1785 1786 1787
            input_layer = self.get_input_layer(i * 3)
            config_assert(input_layer.size == 1, (
                "Inputs for this layer are made up of "
                "several triples, in which the first one is scores over "
                "all candidate paths, whose size should be equal to 1."))
C
caoying03 已提交
1788 1789


Z
zhangjinchao01 已提交
1790 1791
@config_layer('fc')
class FCLayer(LayerBase):
T
tensor-tang 已提交
1792 1793
    layer_type = 'fc'

L
lianxiaochen 已提交
1794 1795 1796 1797 1798 1799 1800
    def __init__(self,
                 name,
                 size,
                 inputs,
                 bias=True,
                 error_clipping_threshold=None,
                 **xargs):
T
tensor-tang 已提交
1801
        use_mkldnn = bool(int(g_command_config_args.get("use_mkldnn", 0)))
1802 1803
        use_mkldnn_wgt = bool(
            int(g_command_config_args.get("use_mkldnn_wgt", 0)))
T
tensor-tang 已提交
1804 1805 1806 1807 1808 1809 1810
        if use_mkldnn:
            self.layer_type = 'mkldnn_fc'
            config_assert(
                len(inputs) == 1,
                "MkldnnFCLayer support one and only one input!")
        super(FCLayer, self).__init__(
            name, self.layer_type, size, inputs=inputs, **xargs)
Z
zhangjinchao01 已提交
1811 1812 1813 1814 1815 1816
        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
            psize = self.config.size * input_layer.size
            dims = [input_layer.size, self.config.size]
            format = self.inputs[input_index].format
            sparse = format == "csr" or format == "csc"
T
tensor-tang 已提交
1817 1818 1819
            if use_mkldnn:
                config_assert(not sparse,
                              "MkldnnFCLayer do not support sparse format yet")
T
tensor-tang 已提交
1820 1821
                if use_mkldnn_wgt:
                    dims = [self.config.size, input_layer.size]
Z
zhangjinchao01 已提交
1822 1823
            if sparse:
                psize = self.inputs[input_index].nnz
1824 1825
            else:
                sparse = None
Z
zhangjinchao01 已提交
1826

Q
qijun 已提交
1827 1828
            self.create_input_parameter(input_index, psize, dims, sparse,
                                        format)
Z
zhangjinchao01 已提交
1829
        self.create_bias_parameter(bias, self.config.size)
L
lianxiaochen 已提交
1830 1831
        if error_clipping_threshold is not None:
            self.config.error_clipping_threshold = error_clipping_threshold
Z
zhangjinchao01 已提交
1832

Q
qijun 已提交
1833

T
tensor-tang 已提交
1834 1835 1836 1837 1838
@config_layer('mkldnn_fc')
class MkldnnFcLayer(FCLayer):
    layer_type = 'mkldnn_fc'


Z
zhangjinchao01 已提交
1839 1840
@config_layer('selective_fc')
class SelectiveFCLayer(LayerBase):
Q
qijun 已提交
1841 1842 1843 1844 1845 1846 1847 1848 1849 1850
    def __init__(self,
                 name,
                 size,
                 inputs,
                 bias=True,
                 selective_fc_pass_generation=False,
                 has_selected_colums=True,
                 selective_fc_full_mul_ratio=0.02,
                 selective_fc_parallel_plain_mul_thread_num=None,
                 **xargs):
Z
zhangjinchao01 已提交
1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870
        super(SelectiveFCLayer, self).__init__(
            name, 'selective_fc', size, inputs=inputs, **xargs)
        # user MUST know if selctive fc is used in training,
        # parameter matrices saved by this layer are automatically transposed,
        # BUT bias is not.

        # if selective_fc is used only in testing mode, and parameters for
        # this layer are trained by fully connected layers,
        # then TranposedFullMatrixProjectin MUST be used in training
        # to avoid manual transpose in testing.

        self.config.selective_fc_pass_generation = selective_fc_pass_generation
        self.config.has_selected_colums = has_selected_colums
        self.config.selective_fc_full_mul_ratio = selective_fc_full_mul_ratio
        if selective_fc_parallel_plain_mul_thread_num is not None:
            self.config.selective_fc_parallel_plain_mul_thread_num = selective_fc_parallel_plain_mul_thread_num

        input_num = len(self.inputs)
        if has_selected_colums:
            config_assert(input_num >= 2,
Q
qijun 已提交
1871 1872
                          ("if indices of selected columns are not specified, "
                           "selective_fc Layer has at least two inputs"))
Z
zhangjinchao01 已提交
1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884
            input_num -= 1

        for input_index in xrange(input_num):
            input_layer = self.get_input_layer(input_index)
            psize = self.config.size * input_layer.size
            dims = [input_layer.size, self.config.size]
            dims = dims[::-1]  # transpose the parameter
            format = self.inputs[input_index].format
            sparse = format == "csr" or format == "csc"
            if sparse:
                psize = self.inputs[input_index].nnz

Q
qijun 已提交
1885 1886
            self.create_input_parameter(input_index, psize, dims, sparse,
                                        format)
Z
zhangjinchao01 已提交
1887 1888
        self.create_bias_parameter(bias, self.config.size)

Q
qijun 已提交
1889

1890 1891
@config_layer('print')
class PrintLayer(LayerBase):
1892
    def __init__(self, name, inputs, format=None):
1893
        super(PrintLayer, self).__init__(name, 'print', 0, inputs)
1894 1895 1896 1897 1898 1899
        if format is None:
            format = "\n".join([
                "layer=" + input.input_layer_name + " %s"
                for input in self.inputs
            ])
        self.config.user_arg = format
1900

Q
qijun 已提交
1901

Y
yuan 已提交
1902 1903
@config_layer('priorbox')
class PriorBoxLayer(LayerBase):
G
gaoyuan 已提交
1904 1905
    def __init__(self, name, inputs, size, min_size, max_size, aspect_ratio,
                 variance):
Y
yuan 已提交
1906
        super(PriorBoxLayer, self).__init__(name, 'priorbox', 0, inputs)
G
gaoyuan 已提交
1907
        config_assert(len(inputs) == 2, 'PriorBoxLayer must have 2 inputs')
G
gaoyuan 已提交
1908 1909 1910 1911 1912 1913 1914
        input_layer = self.get_input_layer(1)
        config_assert(
            input_layer.type == 'data',
            'Expecting the second input layer of an priorbox layer to be '
            'a data layer')
        config_assert(input_layer.width > 0, 'The data layer must set width')
        config_assert(input_layer.height > 0, 'The data layer must set height')
G
gaoyuan 已提交
1915
        config_assert(len(variance) == 4, 'The variance must have 4 inputs')
Y
yuan 已提交
1916 1917 1918 1919 1920 1921
        self.config.inputs[0].priorbox_conf.min_size.extend(min_size)
        self.config.inputs[0].priorbox_conf.max_size.extend(max_size)
        self.config.inputs[0].priorbox_conf.aspect_ratio.extend(aspect_ratio)
        self.config.inputs[0].priorbox_conf.variance.extend(variance)
        self.config.size = size

Q
qijun 已提交
1922

1923 1924 1925
@config_layer('multibox_loss')
class MultiBoxLossLayer(LayerBase):
    def __init__(self, name, inputs, input_num, num_classes, overlap_threshold,
1926
                 neg_pos_ratio, neg_overlap, background_id, **xargs):
1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947
        super(MultiBoxLossLayer, self).__init__(name, 'multibox_loss', 0,
                                                inputs)
        config_assert(
            len(inputs) == (input_num * 2 + 2),
            'MultiBoxLossLayer does not have enough inputs')
        config_assert(num_classes > background_id,
                      'Classes number must greater than background ID')
        self.config.inputs[0].multibox_loss_conf.num_classes = num_classes
        self.config.inputs[
            0].multibox_loss_conf.overlap_threshold = overlap_threshold
        self.config.inputs[0].multibox_loss_conf.neg_pos_ratio = neg_pos_ratio
        self.config.inputs[0].multibox_loss_conf.neg_overlap = neg_overlap
        self.config.inputs[0].multibox_loss_conf.background_id = background_id
        self.config.inputs[0].multibox_loss_conf.input_num = input_num
        self.config.size = 1


@config_layer('detection_output')
class DetectionOutputLayer(LayerBase):
    def __init__(self, name, inputs, size, input_num, num_classes,
                 nms_threshold, nms_top_k, keep_top_k, confidence_threshold,
1948
                 background_id, **xargs):
1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968
        super(DetectionOutputLayer, self).__init__(name, 'detection_output', 0,
                                                   inputs)
        config_assert(
            len(inputs) == (input_num * 2 + 1),
            'DetectionOutputLayer does not have enough inputs')
        config_assert(num_classes > background_id,
                      'Classes number must greater than background ID')
        self.config.inputs[0].detection_output_conf.num_classes = num_classes
        self.config.inputs[
            0].detection_output_conf.nms_threshold = nms_threshold
        self.config.inputs[0].detection_output_conf.nms_top_k = nms_top_k
        self.config.inputs[0].detection_output_conf.keep_top_k = keep_top_k
        self.config.inputs[
            0].detection_output_conf.confidence_threshold = confidence_threshold
        self.config.inputs[
            0].detection_output_conf.background_id = background_id
        self.config.inputs[0].detection_output_conf.input_num = input_num
        self.config.size = size


Z
zhangjinchao01 已提交
1969 1970
@config_layer('data')
class DataLayer(LayerBase):
C
chengduoZH 已提交
1971 1972 1973 1974 1975 1976 1977
    def __init__(self,
                 name,
                 size,
                 depth=None,
                 height=None,
                 width=None,
                 device=None):
Q
qijun 已提交
1978 1979
        super(DataLayer, self).__init__(
            name, 'data', size, inputs=[], device=device)
L
Luo Tao 已提交
1980 1981
        if height and width:
            self.set_layer_height_width(height, width)
C
chengduoZH 已提交
1982 1983
        if depth:
            self.set_layer_depth(depth)
Q
qijun 已提交
1984

Z
zhangjinchao01 已提交
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

'''
DataNormLayer: A layer for data normalization
Input: One and only one input layer is accepted. The input layer must
       be DataLayer with dense data type
Output: The normalization of the input data

Reference:
    LA Shalabi, Z Shaaban, B Kasasbeh. Data mining: A preprocessing engine

Example:
    Layer(
        name = "norm_input_layer",
        type = "data_norm",
        inputs = [Input("input_layer",
                        parameter_name = "_slot0.stats")],
        data_norm_strategy = "z-score",
    )

Note:
  (1) The parameter has been calculated in the preprocessing stage,
      and should be initialized by --init_model_path when training.
  (2) Three data normalization methoeds are considered
          z-score: y = (x-mean)/std
          min-max: y = (x-min)/(max-min)
          decimal-scaling: y = x/10^j, where j is the smallest integer such that max(|y|)<1
'''
Q
qijun 已提交
2012 2013


Z
zhangjinchao01 已提交
2014 2015
@config_layer('data_norm')
class DataNormLayer(LayerBase):
Q
qijun 已提交
2016
    def __init__(self, name, inputs, data_norm_strategy="z-score", device=None):
Z
zhangjinchao01 已提交
2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027
        super(DataNormLayer, self).__init__(
            name, 'data_norm', 0, inputs=inputs, device=device)
        self.config.data_norm_strategy = data_norm_strategy
        config_assert(len(inputs) == 1, 'DataNormLayer must have 1 input')
        input_layer = self.get_input_layer(0)
        self.set_layer_size(input_layer.size)
        para_size = 5 * input_layer.size
        para_dims = [5, input_layer.size]
        self.inputs[0].is_static = True
        self.create_input_parameter(0, para_size, para_dims)

Q
qijun 已提交
2028

Z
zhangjinchao01 已提交
2029 2030 2031
@config_layer('prelu')
class ParameterReluLayer(LayerBase):
    layer_type = 'prelu'
Q
qijun 已提交
2032 2033

    def __init__(self, name, inputs, partial_sum=1, **args):
Z
zhangjinchao01 已提交
2034 2035 2036
        super(ParameterReluLayer, self).__init__(
            name, self.layer_type, 0, inputs=inputs, **args)
        input_layer = self.get_input_layer(0)
2037 2038 2039
        config_assert(len(self.inputs) == 1, "prelu layer has only one input.")
        config_assert(input_layer.size % partial_sum == 0,
                      "a wrong setting for partial_sum")
Z
zhangjinchao01 已提交
2040
        self.set_layer_size(input_layer.size)
C
caoying03 已提交
2041
        self.config.partial_sum = partial_sum
Z
zhangjinchao01 已提交
2042 2043
        self.create_input_parameter(0, input_layer.size / partial_sum)

Q
qijun 已提交
2044

Z
zhangjinchao01 已提交
2045 2046 2047
@config_layer('conv')
class ConvLayerBase(LayerBase):
    layer_type = 'conv'
Q
qijun 已提交
2048 2049 2050 2051 2052 2053 2054 2055

    def __init__(self,
                 name,
                 inputs=[],
                 bias=True,
                 num_filters=None,
                 shared_biases=False,
                 **xargs):
Z
zhangjinchao01 已提交
2056 2057 2058 2059 2060 2061
        super(ConvLayerBase, self).__init__(
            name, self.layer_type, 0, inputs=inputs, **xargs)

        if num_filters is not None:
            self.config.num_filters = num_filters

2062
        use_mkldnn = int(g_command_config_args.get("use_mkldnn", 0))
Z
zhangjinchao01 已提交
2063 2064 2065
        use_gpu = int(g_command_config_args.get("use_gpu", 0))
        parallel_nn = int(g_command_config_args.get("parallel_nn", 0))

2066 2067
        # Automatically select cudnn_type for GPU, exconv for CPU
        # and mkldnn_conv for MKLDNN
Z
zhangjinchao01 已提交
2068
        # if set type=conv, but still reserve the way user specify
2069
        # exconv, mkldnn_conv or cudnn_conv manually.
Z
zhangjinchao01 已提交
2070 2071 2072
        if self.layer_type == "cudnn_conv":
            config_assert(use_gpu, "cudnn_conv only support GPU")

2073 2074 2075
        if self.layer_type == "mkldnn_conv":
            config_assert(use_mkldnn, "mkldnn_conv only support MKLDNN")

Z
zhangjinchao01 已提交
2076
        if (use_gpu == 1 and self.layer_type != "exconv" and
2077
                self.layer_type != "mkldnn_conv" and
Q
qijun 已提交
2078
            (parallel_nn == 0 or self.config.device > -1)):
Z
zhangjinchao01 已提交
2079 2080
            self.layer_type = "cudnn_conv"
        else:
2081
            self.layer_type = "mkldnn_conv" if use_mkldnn else "exconv"
Z
zhangjinchao01 已提交
2082 2083 2084 2085 2086 2087 2088 2089 2090
        # need to specify layer in config
        self.config.type = self.layer_type

        if shared_biases is not None:
            self.config.shared_biases = shared_biases

        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
            conv_conf = self.config.inputs[input_index].conv_conf
L
Luo Tao 已提交
2091 2092
            parse_conv(self.inputs[input_index].conv, input_layer.name,
                       conv_conf, num_filters)
Z
zhangjinchao01 已提交
2093 2094
            psize = self.calc_parameter_size(conv_conf)
            self.create_input_parameter(input_index, psize)
L
Luo Tao 已提交
2095 2096
            self.set_cnn_layer(name, conv_conf.output_y, conv_conf.output_x,
                               self.config.num_filters)
Z
zhangjinchao01 已提交
2097 2098 2099 2100 2101 2102 2103 2104

        psize = self.config.size
        if shared_biases:
            psize = self.config.num_filters
        self.create_bias_parameter(bias, psize, [psize, 1])

    def calc_parameter_size(self, conv_conf):
        return self.config.num_filters * conv_conf.filter_channels \
2105
               * (conv_conf.filter_size * conv_conf.filter_size_y)
Z
zhangjinchao01 已提交
2106

Q
qijun 已提交
2107

Z
zhangjinchao01 已提交
2108 2109 2110 2111
@config_layer('exconv')
class ConvLayer(ConvLayerBase):
    layer_type = 'exconv'

Q
qijun 已提交
2112

2113 2114 2115 2116 2117
@config_layer('mkldnn_conv')
class ConvLayer(ConvLayerBase):
    layer_type = 'mkldnn_conv'


Z
zhangjinchao01 已提交
2118 2119 2120 2121
@config_layer('cudnn_conv')
class ConvLayer(ConvLayerBase):
    layer_type = 'cudnn_conv'

2122 2123 2124 2125

@config_layer('convt')
class ConvTransLayerBase(LayerBase):
    layer_type = 'convt'
Q
qijun 已提交
2126 2127 2128 2129 2130 2131 2132 2133

    def __init__(self,
                 name,
                 inputs=[],
                 bias=True,
                 num_filters=None,
                 shared_biases=False,
                 **xargs):
2134
        super(ConvTransLayerBase, self).__init__(
2135 2136 2137 2138 2139 2140 2141 2142
            name, self.layer_type, 0, inputs=inputs, **xargs)

        if num_filters is not None:
            self.config.num_filters = num_filters

        use_gpu = int(g_command_config_args.get("use_gpu", 0))
        parallel_nn = int(g_command_config_args.get("parallel_nn", 0))

2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153
        # Automatically select cudnn_type for GPU and exconvt for CPU
        # if set type=exconvt, but still reserve the way user specify
        # exconvt or cudnn_convt manually.
        if self.layer_type == "cudnn_convt":
            config_assert(use_gpu, "cudnn_convt only support GPU")

        if (use_gpu == 1 and self.layer_type != "exconvt" and
            (parallel_nn == 0 or self.config.device > -1)):
            self.layer_type = "cudnn_convt"
        else:
            self.layer_type = "exconvt"
2154 2155 2156 2157 2158 2159 2160 2161
        # need to specify layer in config
        self.config.type = self.layer_type

        if shared_biases is not None:
            self.config.shared_biases = shared_biases

        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
2162
            parse_conv(
2163 2164
                self.inputs[input_index].conv,
                input_layer.name,
2165
                self.config.inputs[input_index].conv_conf,
2166
                num_filters,
2167
                trans=True)
2168 2169 2170
            conv_conf = self.config.inputs[input_index].conv_conf
            psize = self.calc_parameter_size(conv_conf)
            self.create_input_parameter(input_index, psize)
2171 2172
            self.set_cnn_layer(name, conv_conf.img_size_y, conv_conf.img_size,
                               self.config.num_filters)
2173 2174 2175 2176 2177 2178 2179

        psize = self.config.size
        if shared_biases:
            psize = self.config.num_filters
        self.create_bias_parameter(bias, psize, [psize, 1])

    def calc_parameter_size(self, conv_conf):
2180
        return conv_conf.channels * conv_conf.filter_channels \
2181 2182
                    * (conv_conf.filter_size * conv_conf.filter_size_y)

Q
qijun 已提交
2183

2184 2185 2186 2187
@config_layer('exconvt')
class ConvTransLayer(ConvTransLayerBase):
    layer_type = 'exconvt'

Q
qijun 已提交
2188

2189 2190 2191 2192 2193
@config_layer('cudnn_convt')
class ConvTransLayer(ConvTransLayerBase):
    layer_type = 'cudnn_convt'


C
chengduoZH 已提交
2194 2195
@config_layer('conv_3d')
class Conv3DLayerBase(LayerBase):
2196 2197 2198 2199 2200
    def __init__(self,
                 name,
                 inputs=[],
                 bias=True,
                 num_filters=None,
C
chengduoZH 已提交
2201
                 shared_biases=True,
2202
                 **xargs):
C
chengduoZH 已提交
2203
        super(Conv3DLayerBase, self).__init__(
2204 2205 2206 2207 2208 2209 2210 2211
            name, self.layer_type, 0, inputs=inputs, **xargs)

        if num_filters is not None:
            self.config.num_filters = num_filters

        # need to specify layer in config
        self.config.type = self.layer_type

C
chengduoZH 已提交
2212 2213 2214 2215
        trans = False
        if self.config.type == "deconv3d":
            trans = True

2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226
        if shared_biases is not None:
            self.config.shared_biases = shared_biases

        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
            conv_conf = self.config.inputs[input_index].conv_conf
            parse_conv3d(
                self.inputs[input_index].conv,
                input_layer.name,
                conv_conf,
                num_filters,
C
chengduoZH 已提交
2227
                trans=trans
2228 2229 2230
            )  # for z-axis pad:0, strid:1, filter_size:1, img_size:1
            psize = self.calc_parameter_size(conv_conf)
            self.create_input_parameter(input_index, psize)
C
chengduoZH 已提交
2231 2232 2233 2234 2235 2236 2237
            if trans:
                self.set_cnn_layer(name, conv_conf.img_size_z,
                                   conv_conf.img_size_y, conv_conf.img_size,
                                   self.config.num_filters)
            else:
                self.set_cnn_layer(name, conv_conf.output_z, conv_conf.output_y,
                                   conv_conf.output_x, self.config.num_filters)
2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257

        psize = self.config.size
        if shared_biases:
            psize = self.config.num_filters
        self.create_bias_parameter(bias, psize, [psize, 1])

    def calc_parameter_size(self, conv_conf):
        return self.config.num_filters * conv_conf.filter_channels \
               * (conv_conf.filter_size * conv_conf.filter_size_y \
                  * conv_conf.filter_size_z)

    def set_cnn_layer(self,
                      input_layer_name,
                      depth,
                      height,
                      width,
                      channels,
                      is_print=True):
        size = depth * height * width * channels
        self.set_layer_size(size)
C
chengduoZH 已提交
2258 2259
        self.set_layer_height_width(height, width)
        self.set_layer_depth(depth)
2260 2261 2262 2263 2264
        if is_print:
            print("output for %s: c = %d, d = %d, h = %d, w = %d, size = %d" %
                  (input_layer_name, channels, depth, height, width, size))


C
chengduoZH 已提交
2265 2266 2267
@config_layer('conv3d')
class Conv3DLayer(Conv3DLayerBase):
    layer_type = 'conv3d'
2268

Q
qijun 已提交
2269

C
chengduoZH 已提交
2270 2271 2272
@config_layer('deconv3d')
class Conv3DLayer(Conv3DLayerBase):
    layer_type = 'deconv3d'
2273 2274


Z
zhangjinchao01 已提交
2275 2276
@config_layer('norm')
class NormLayer(LayerBase):
2277 2278
    def __init__(self, name, inputs, **xargs):
        super(NormLayer, self).__init__(name, 'norm', 0, inputs=inputs, **xargs)
Z
zhangjinchao01 已提交
2279 2280 2281
        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
            norm_conf = self.config.inputs[input_index].norm_conf
L
Luo Tao 已提交
2282 2283 2284 2285
            parse_norm(self.inputs[input_index].norm, input_layer.name,
                       norm_conf)
            self.set_cnn_layer(name, norm_conf.output_y, norm_conf.output_x,
                               norm_conf.channels, False)
2286 2287 2288
            if norm_conf.norm_type == "cross-channel-norm":
                self.create_input_parameter(0, norm_conf.channels,
                                            [norm_conf.channels, 1])
Q
qijun 已提交
2289

Z
zhangjinchao01 已提交
2290 2291 2292

@config_layer('pool')
class PoolLayer(LayerBase):
2293 2294
    layer_type = 'pool'

2295
    def __init__(self, name, inputs, ceil_mode=True, **xargs):
2296 2297 2298 2299 2300 2301
        use_mkldnn = int(g_command_config_args.get("use_mkldnn", 0))
        if self.layer_type == "mkldnn_pool":
            config_assert(use_mkldnn, "mkldnn_pool only support MKLDNN")
        self.layer_type = 'mkldnn_pool' if use_mkldnn else 'pool'
        super(PoolLayer, self).__init__(
            name, self.layer_type, 0, inputs=inputs, **xargs)
Z
zhangjinchao01 已提交
2302 2303 2304
        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
            pool_conf = self.config.inputs[input_index].pool_conf
L
Luo Tao 已提交
2305
            parse_pool(self.inputs[input_index].pool, input_layer.name,
2306
                       pool_conf, ceil_mode)
L
Luo Tao 已提交
2307 2308
            self.set_cnn_layer(name, pool_conf.output_y, pool_conf.output_x,
                               pool_conf.channels)
Q
qijun 已提交
2309

Z
zhangjinchao01 已提交
2310

2311 2312 2313 2314 2315
@config_layer('mkldnn_pool')
class MKLDNNPoolLayer(PoolLayer):
    layer_type = 'mkldnn_pool'


C
chengduoZH 已提交
2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344
@config_layer('pool3d')
class Pool3DLayer(LayerBase):
    def __init__(self, name, inputs, ceil_mode=True, **xargs):
        super(Pool3DLayer, self).__init__(
            name, 'pool3d', 0, inputs=inputs, **xargs)
        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
            pool_conf = self.config.inputs[input_index].pool_conf
            parse_pool3d(self.inputs[input_index].pool, input_layer.name,
                         pool_conf, ceil_mode)
            self.set_cnn_layer(name, pool_conf.output_z, pool_conf.output_y,
                               pool_conf.output_x, pool_conf.channels)

    def set_cnn_layer(self,
                      input_layer_name,
                      depth,
                      height,
                      width,
                      channels,
                      is_print=True):
        size = depth * height * width * channels
        self.set_layer_size(size)
        self.set_layer_height_width(height, width)
        self.set_layer_depth(depth)
        if is_print:
            print("output for %s: c = %d, d = %d, h = %d, w = %d, size = %d" %
                  (input_layer_name, channels, depth, height, width, size))


Q
qijun 已提交
2345 2346
@config_layer('spp')
class SpatialPyramidPoolLayer(LayerBase):
2347
    def __init__(self, name, inputs, **xargs):
Q
qijun 已提交
2348
        super(SpatialPyramidPoolLayer, self).__init__(
2349
            name, 'spp', 0, inputs=inputs, **xargs)
Q
qijun 已提交
2350 2351 2352
        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
            spp_conf = self.config.inputs[input_index].spp_conf
L
Luo Tao 已提交
2353 2354 2355
            parse_spp(self.inputs[input_index].spp, input_layer.name, spp_conf)
            output_x = (pow(4, spp_conf.pyramid_height) - 1) / (4 - 1)
            self.set_cnn_layer(name, 1, output_x, spp_conf.image_conf.channels)
Q
qijun 已提交
2356

Q
qijun 已提交
2357

D
dangqingqing 已提交
2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376
@config_layer('pad')
class PadLayer(LayerBase):
    def __init__(self, name, inputs, **xargs):
        super(PadLayer, self).__init__(name, 'pad', 0, inputs=inputs, **xargs)
        pad = self.inputs[0].pad
        self.config.inputs[0].pad_conf.pad_c.extend(pad.pad_c)
        self.config.inputs[0].pad_conf.pad_h.extend(pad.pad_h)
        self.config.inputs[0].pad_conf.pad_w.extend(pad.pad_w)

        input_layer = self.get_input_layer(0)
        image_conf = self.config.inputs[0].pad_conf.image_conf
        parse_image(pad, input_layer.name, image_conf)
        out_ch = pad.channels + pad.pad_c[0] + pad.pad_c[1]
        out_h = image_conf.img_size_y + pad.pad_h[0] + pad.pad_h[1]
        out_w = image_conf.img_size + pad.pad_w[0] + pad.pad_w[1]
        self.set_cnn_layer(name, out_h, out_w, out_ch)
        self.config.size = out_ch * out_h * out_w


2377 2378
@config_layer('crop')
class CropLayer(LayerBase):
2379
    def __init__(self, name, inputs, axis, offset, shape, **xargs):
2380
        super(CropLayer, self).__init__(name, 'crop', 0, inputs=inputs, **xargs)
2381 2382 2383
        self.config.axis = axis
        self.config.offset.extend(offset)
        self.config.shape.extend(shape)
2384 2385 2386 2387 2388 2389 2390 2391 2392 2393

        # get channel, width and height from input_0 layer
        input_layer = self.get_input_layer(0)
        image_conf = self.config.inputs[0].image_conf
        image_conf.img_size = input_layer.width
        image_conf.img_size_y = input_layer.height
        image_conf.channels = input_layer.size / (input_layer.width *
                                                  input_layer.height)


Z
zhangjinchao01 已提交
2394 2395 2396
@config_layer('batch_norm')
class BatchNormLayer(LayerBase):
    layer_type = 'batch_norm'
Q
qijun 已提交
2397 2398 2399 2400 2401

    def __init__(self,
                 name,
                 inputs,
                 bias=True,
C
chengduoZH 已提交
2402
                 img3D=False,
Q
qijun 已提交
2403 2404 2405
                 use_global_stats=True,
                 moving_average_fraction=0.9,
                 batch_norm_type=None,
C
chengduoZH 已提交
2406
                 mean_var_names=None,
Q
qijun 已提交
2407
                 **xargs):
Z
zhangjinchao01 已提交
2408 2409 2410 2411
        if inputs is None:
            inputs = []
        elif not isinstance(inputs, list):
            inputs = [inputs]
Q
qijun 已提交
2412 2413
        config_assert(
            len(inputs) == 1, "BatchNormLayer must have one and only one input")
Z
zhangjinchao01 已提交
2414 2415 2416 2417 2418 2419 2420 2421
        # Create Input for moving mean and std,
        # in batch normalization layer.
        # These paras no need to update, so set is_static is true.
        # If not use is_static, even set learning_rate = 0, decay_rate = 0,
        # these paras will change if set average_window in configure.
        use_gpu = bool(int(g_command_config_args.get("use_gpu", 0)))
        is_shared = True if not use_gpu else False
        for i in xrange(2):
Q
qijun 已提交
2422 2423 2424 2425 2426 2427
            inputs.append(
                Input(
                    inputs[0].input_layer_name,
                    initial_std=0.0,
                    initial_mean=0.0,
                    is_static=True,
2428
                    is_shared=is_shared,
D
dangqingqing 已提交
2429
                    make_layer_name_in_submodel=False, ))
Z
zhangjinchao01 已提交
2430 2431 2432 2433 2434 2435

        parallel_nn = bool(int(g_command_config_args.get("parallel_nn", 0)))
        cudnn_version = int(g_command_config_args.get("cudnn_version", 0))
        # Automatically select cudnn_batch_norm for GPU and batch_norm for CPU.
        # Also based on cudnn version.
        use_cudnn = use_gpu and batch_norm_type != "batch_norm" and \
2436
                ((not parallel_nn) or self.config.device > -1)
Z
zhangjinchao01 已提交
2437
        self.layer_type = "cudnn_batch_norm" if use_cudnn else "batch_norm"
Q
qijun 已提交
2438
        super(BatchNormLayer, self).__init__(
X
xuwei06 已提交
2439
            name, self.layer_type, 0, inputs=inputs, **xargs)
Z
zhangjinchao01 已提交
2440 2441 2442 2443 2444 2445

        if use_global_stats is not None:
            self.config.use_global_stats = use_global_stats
        if moving_average_fraction is not None:
            self.config.moving_average_fraction = moving_average_fraction

Q
qijun 已提交
2446
        input_layer = self.get_input_layer(0)
Z
zhangjinchao01 已提交
2447
        image_conf = self.config.inputs[0].image_conf
C
chengduoZH 已提交
2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461
        if img3D:
            parse_image3d(self.inputs[0].image, input_layer.name, image_conf)
            # Only pass the width and height of input to batch_norm layer
            # when either of it is non-zero.
            if input_layer.width != 0 or input_layer.height != 0:
                self.set_cnn_layer(
                    input_layer_name=name,
                    depth=image_conf.img_size_z,
                    height=image_conf.img_size_y,
                    width=image_conf.img_size,
                    channels=image_conf.channels,
                    is_print=True)
            else:
                self.set_layer_size(input_layer.size)
2462
        else:
C
chengduoZH 已提交
2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474
            parse_image(self.inputs[0].image, input_layer.name, image_conf)
            # Only pass the width and height of input to batch_norm layer
            # when either of it is non-zero.
            if input_layer.width != 0 or input_layer.height != 0:
                self.set_cnn_layer(
                    input_layer_name=name,
                    height=image_conf.img_size_y,
                    width=image_conf.img_size,
                    channels=image_conf.channels,
                    is_print=True)
            else:
                self.set_layer_size(input_layer.size)
Z
zhangjinchao01 已提交
2475 2476 2477

        psize = self.calc_parameter_size(image_conf)
        dims = [1, psize]
C
chengduoZH 已提交
2478 2479 2480 2481
        if mean_var_names is not None:
            assert len(mean_var_names) == 2
            self.inputs[1].parameter_name = mean_var_names[0]
            self.inputs[2].parameter_name = mean_var_names[1]
C
chengduoZH 已提交
2482

Z
zhangjinchao01 已提交
2483 2484 2485 2486 2487 2488
        self.create_input_parameter(0, psize)
        self.create_input_parameter(1, psize, dims)
        self.create_input_parameter(2, psize, dims)

        self.create_bias_parameter(bias, psize)

C
chengduoZH 已提交
2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510
    def set_cnn_layer(self,
                      input_layer_name,
                      depth=None,
                      height=None,
                      width=None,
                      channels=None,
                      is_print=True):
        depthIsNone = False
        if depth is None:
            depth = 1
            depthIsNone = True
        size = depth * height * width * channels
        self.set_layer_size(size)
        self.set_layer_height_width(height, width)
        self.set_layer_depth(depth)
        if is_print and depthIsNone:
            print("output for %s: c = %d, h = %d, w = %d, size = %d" %
                  (input_layer_name, channels, height, width, size))
        elif is_print:
            print("output for %s: c = %d, d = %d, h = %d, w = %d, size = %d" %
                  (input_layer_name, channels, depth, height, width, size))

Z
zhangjinchao01 已提交
2511 2512 2513
    def calc_parameter_size(self, image_conf):
        return image_conf.channels

Q
qijun 已提交
2514

Z
zhangjinchao01 已提交
2515 2516
@config_layer('trans')
class TransLayer(LayerBase):
2517
    def __init__(self, name, inputs, **xargs):
Q
qijun 已提交
2518
        super(TransLayer, self).__init__(
2519
            name, 'trans', 0, inputs=inputs, **xargs)
Q
qijun 已提交
2520 2521 2522
        config_assert(
            len(self.inputs) == 1,
            'TransLayer must have one and only one input')
Z
zhangjinchao01 已提交
2523 2524
        self.set_layer_size(self.get_input_layer(0).size)

Q
qijun 已提交
2525

Z
zhangjinchao01 已提交
2526 2527
@config_layer('resize')
class ResizeLayer(LayerBase):
2528
    def __init__(self, name, size, inputs, **xargs):
Q
qijun 已提交
2529
        super(ResizeLayer, self).__init__(
2530
            name, 'resize', size=size, inputs=inputs, **xargs)
Q
qijun 已提交
2531 2532 2533 2534
        config_assert(
            len(self.inputs) == 1,
            'ResizeLayer must have one and only one input')

Z
zhangjinchao01 已提交
2535

2536 2537
@config_layer('rotate')
class RotateLayer(LayerBase):
H
Haonan 已提交
2538
    def __init__(self, name, inputs, height, width, device=None):
2539 2540 2541 2542 2543
        super(RotateLayer, self).__init__(
            name, 'rotate', 0, inputs=inputs, device=device)
        config_assert(
            len(self.inputs) == 1,
            'RotateLayer must have one and only one input')
H
Haonan 已提交
2544
        self.set_layer_height_width(height, width)
2545 2546 2547
        self.set_layer_size(self.get_input_layer(0).size)


Z
zhangjinchao01 已提交
2548 2549
@config_layer('blockexpand')
class BlockExpandLayer(LayerBase):
2550
    def __init__(self, name, inputs, **xargs):
Q
qijun 已提交
2551
        super(BlockExpandLayer, self).__init__(
2552
            name, 'blockexpand', 0, inputs=inputs, **xargs)
Z
zhangjinchao01 已提交
2553 2554
        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
Q
qijun 已提交
2555 2556
            parse_block_expand(
                self.inputs[input_index].block_expand, input_layer.name,
Z
zhangjinchao01 已提交
2557
                self.config.inputs[input_index].block_expand_conf)
Q
qijun 已提交
2558 2559 2560 2561 2562 2563
            block_expand_conf = self.config.inputs[
                input_index].block_expand_conf
            self.set_layer_size(block_expand_conf.block_x *
                                block_expand_conf.block_y *
                                block_expand_conf.channels)

Z
zhangjinchao01 已提交
2564

2565 2566
@config_layer('maxout')
class MaxOutLayer(LayerBase):
Q
qijun 已提交
2567 2568 2569
    def __init__(self, name, inputs, **xargs):
        super(MaxOutLayer, self).__init__(
            name, 'maxout', 0, inputs=inputs, **xargs)
2570 2571
        input_layer = self.get_input_layer(0)
        maxout_conf = self.config.inputs[0].maxout_conf
L
Luo Tao 已提交
2572
        parse_maxout(self.inputs[0].maxout, input_layer.name, maxout_conf)
L
Luo Tao 已提交
2573
        out_channels = maxout_conf.image_conf.channels / maxout_conf.groups
2574 2575
        self.set_cnn_layer(name, maxout_conf.image_conf.img_size_y,
                           maxout_conf.image_conf.img_size, out_channels)
Q
qijun 已提交
2576

2577

D
dangqingqing 已提交
2578 2579 2580 2581
@config_layer('row_conv')
class RowConvLayer(LayerBase):
    def __init__(self, name, inputs, context_length, **xargs):
        super(RowConvLayer, self).__init__(
2582
            name, 'row_conv', 0, inputs=inputs, **xargs)
D
dangqingqing 已提交
2583 2584
        config_assert(
            len(self.inputs) == 1,
2585
            'row convolution layer must have one and only one input.')
D
dangqingqing 已提交
2586 2587 2588 2589 2590 2591 2592 2593 2594
        input_layer = self.get_input_layer(0)
        row_conv_conf = self.config.inputs[0].row_conv_conf
        row_conv_conf.context_length = context_length
        self.set_layer_size(input_layer.size)
        psize = context_length * input_layer.size
        dims = [context_length, input_layer.size]
        self.create_input_parameter(0, psize, dims)


G
guosheng 已提交
2595 2596
@config_layer('clip')
class ClipLayer(LayerBase):
2597 2598
    def __init__(self, name, inputs, min, max, **xargs):
        super(ClipLayer, self).__init__(name, 'clip', 0, inputs=inputs, **xargs)
G
guosheng 已提交
2599 2600
        config_assert(
            len(self.inputs) == 1,
2601 2602
            'ClipLayer must have one and only one input.')
        config_assert(min < max, 'min must be less than max.')
G
guosheng 已提交
2603 2604
        input_layer = self.get_input_layer(0)
        self.set_layer_size(input_layer.size)
2605 2606
        self.config.inputs[0].clip_conf.min = min
        self.config.inputs[0].clip_conf.max = max
G
guosheng 已提交
2607 2608


G
guosheng 已提交
2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622
@config_layer('scale_shift')
class ScaleShiftLayer(LayerBase):
    def __init__(self, name, inputs, bias=True, **xargs):
        super(ScaleShiftLayer, self).__init__(
            name, 'scale_shift', 0, inputs=inputs, **xargs)
        config_assert(
            len(self.inputs) == 1,
            'ScaleShiftLayer must have one and only one input.')
        input_layer = self.get_input_layer(0)
        self.set_layer_size(input_layer.size)
        self.create_input_parameter(0, 1, [1, 1])
        self.create_bias_parameter(bias, 1)


Z
zhangjinchao01 已提交
2623 2624 2625 2626
# key: cost type
# value: cost class
g_cost_map = {}

Q
qijun 已提交
2627

Z
zhangjinchao01 已提交
2628 2629 2630
# define a cost layer without any parameters
def define_cost(class_name, cost_type):
    def init(cls, name, inputs, device=None, coeff=1.):
Q
qijun 已提交
2631 2632
        super(type(cls), cls).__init__(
            name, cost_type, 1, inputs, device=device, coeff=coeff)
Z
zhangjinchao01 已提交
2633

Q
qijun 已提交
2634
    cls = type(class_name, (LayerBase, ), dict(__init__=init))
Z
zhangjinchao01 已提交
2635 2636 2637
    global g_cost_map
    g_cost_map[cost_type] = cls

Q
qijun 已提交
2638

Z
zhangjinchao01 已提交
2639
define_cost('MultiClassCrossEntropy', 'multi-class-cross-entropy')
C
caoying03 已提交
2640
define_cost('CrossEntropyOverBeamCostLayer', 'cross_entropy_over_beam')
Z
zhangjinchao01 已提交
2641 2642 2643 2644 2645 2646
define_cost('RankingCost', 'rank-cost')
define_cost('AucValidation', 'auc-validation')
define_cost('PnpairValidation', 'pnpair-validation')
define_cost('SumOfSquaresCostLayer', 'square_error')
define_cost('MultiBinaryLabelCrossEntropy', 'multi_binary_label_cross_entropy')
define_cost('SoftBinaryClassCrossEntropy', 'soft_binary_class_cross_entropy')
2647
define_cost('HuberTwoClassification', 'huber_classification')
X
xuwei06 已提交
2648
define_cost('SumCost', 'sum_cost')
D
dangqingqing 已提交
2649
define_cost('SmoothL1Cost', 'smooth_l1')
Z
zhangjinchao01 已提交
2650

Q
qijun 已提交
2651

Z
zhangjinchao01 已提交
2652 2653
@config_layer('hsigmoid')
class HierarchicalSigmoidLayer(LayerBase):
Q
qijun 已提交
2654
    def __init__(self, name, num_classes, inputs, device=None, bias=True):
Z
zhangjinchao01 已提交
2655 2656
        super(HierarchicalSigmoidLayer, self).__init__(
            name, 'hsigmoid', 1, inputs=inputs, device=device)
Q
qijun 已提交
2657 2658 2659
        config_assert(
            len(self.inputs) >= 2,
            'HierarchicalSigmoidLayer must have at least 2 inputs')
Z
zhangjinchao01 已提交
2660 2661 2662 2663 2664 2665 2666 2667
        self.config.num_classes = num_classes
        for input_index in xrange(len(self.inputs) - 1):
            input_layer = self.get_input_layer(input_index)
            psize = (num_classes - 1) * input_layer.size
            dims = [num_classes - 1, input_layer.size]
            self.create_input_parameter(input_index, psize, dims)
        self.create_bias_parameter(bias, num_classes - 1)

Q
qijun 已提交
2668

Z
zhangjinchao01 已提交
2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692
'''
lambdaCost for lambdaRank LTR approach

Usage:
  Example: Layer(name = "cost", type = "lambda_cost", NDCG_num = 8,
             max_sort_size = -1, inputs = ["output", "score"])

  Input data: Samples of the same query should be loaded as a sequence,
          by ProtoDataProvider or PyDataProvider etc.. User should provide
          scores for each sample. The score slot should be the 2nd
          input of lambdaRank layer.

  NDCG_num = the size of NDCG, e.g., 5 for NDCG@5.
    Note: NDCG_num must be less than or equal to the minimum
          size of lists.

  max_sort_size = the size of partial sorting in calculating gradient.
    Note: If max_sort_size = -1, then for each list, the algorithm will
          sort the entire list to get gradient.
          In other cases, max_sort_size must be greater than or equal
          to NDCG_num.
          max_sort_size can be greater than the size of a list, in which
          case the algorithm will sort the entire list to get gradient.
'''
Q
qijun 已提交
2693 2694


Z
zhangjinchao01 已提交
2695 2696
@config_layer('lambda_cost')
class LambdaCost(LayerBase):
Q
qijun 已提交
2697
    def __init__(self, name, inputs, NDCG_num=5, max_sort_size=-1, device=None):
Z
zhangjinchao01 已提交
2698 2699
        super(LambdaCost, self).__init__(
            name, 'lambda_cost', 1, inputs=inputs, device=device)
Q
qijun 已提交
2700
        config_assert(len(self.inputs) == 2, 'lambdaCost must have 2 inputs')
Z
zhangjinchao01 已提交
2701 2702
        self.config.NDCG_num = NDCG_num
        if max_sort_size != -1:
Q
qijun 已提交
2703 2704 2705
            config_assert(
                NDCG_num <= max_sort_size,
                'NDCG_num must be less than or equal to max_sort_size')
Z
zhangjinchao01 已提交
2706 2707
        self.config.max_sort_size = max_sort_size

Q
qijun 已提交
2708

L
Luo Tao 已提交
2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719
@config_layer('huber_regression')
class HuberRegressionLoss(LayerBase):
    def __init__(self, name, inputs, delta=1., coeff=1., device=None):
        super(HuberRegressionLoss, self).__init__(
            name, 'huber_regression', 1, inputs=inputs, device=device)
        config_assert(
            len(self.inputs) == 2, 'HuberRegression must have 2 inputs')
        self.config.delta = delta
        self.config.coeff = coeff


Z
zhangjinchao01 已提交
2720 2721
@config_layer('nce')
class NCELayer(LayerBase):
Q
qijun 已提交
2722 2723 2724 2725 2726 2727 2728 2729
    def __init__(self,
                 name,
                 num_classes,
                 inputs,
                 num_neg_samples=10,
                 neg_sampling_dist=None,
                 bias=True,
                 **xargs):
Z
zhangjinchao01 已提交
2730
        super(NCELayer, self).__init__(name, 'nce', 1, inputs=inputs, **xargs)
Q
qijun 已提交
2731 2732
        config_assert(
            len(self.inputs) >= 2, 'NCELayer must have at least 2 inputs')
Z
zhangjinchao01 已提交
2733 2734
        self.config.num_classes = num_classes
        if neg_sampling_dist is not None:
Q
qijun 已提交
2735 2736 2737 2738
            config_assert(
                len(neg_sampling_dist) == num_classes,
                'len(neg_sampling_dist)(%s) is not same as num_classes (%s)' %
                (len(neg_sampling_dist), num_classes))
Z
zhangjinchao01 已提交
2739
            s = sum(neg_sampling_dist)
Q
qijun 已提交
2740 2741 2742
            config_assert(
                abs(s - 1) < 1e-5,
                'The sum of neg_sampling_dist (%s) is not 1' % s)
Z
zhangjinchao01 已提交
2743 2744 2745 2746 2747

            self.config.neg_sampling_dist.extend(neg_sampling_dist)

        self.config.num_neg_samples = num_neg_samples
        num_real_inputs = len(self.inputs) - 1
Q
qijun 已提交
2748
        input_layer = self.get_input_layer(num_real_inputs)
Z
zhangjinchao01 已提交
2749 2750 2751 2752
        config_assert(input_layer.type == 'data',
                      'Expecting the last input layer of an nce layer to be '
                      'a data layer')

Q
qijun 已提交
2753 2754
        if (num_real_inputs > 1 and input_layer.size == 1 and
                self.get_input_layer(num_real_inputs - 1).type == 'data'):
Z
zhangjinchao01 已提交
2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767
            # This input layer is assumed to be a sample weight layer
            num_real_inputs -= 1

        for input_index in xrange(num_real_inputs):
            input_layer = self.get_input_layer(input_index)
            psize = num_classes * input_layer.size
            dims = [num_classes, input_layer.size]
            self.create_input_parameter(input_index, psize, dims)
        self.create_bias_parameter(bias, num_classes)


@config_layer('addto')
class AddToLayer(LayerBase):
Q
qijun 已提交
2768
    def __init__(self, name, inputs, bias=True, **xargs):
Z
zhangjinchao01 已提交
2769 2770
        super(AddToLayer, self).__init__(
            name, 'addto', 0, inputs=inputs, **xargs)
Q
qijun 已提交
2771
        config_assert(len(inputs) > 0, 'inputs cannot be empty for AddToLayer')
2772 2773

        if len(self.inputs) > 1:
2774 2775 2776 2777 2778 2779 2780
            for input_index in xrange(len(self.inputs)):
                assert self.get_input_layer(0).height == self.get_input_layer(
                    input_index).height
                assert self.get_input_layer(0).width == self.get_input_layer(
                    input_index).width
                assert self.get_input_layer(0).depth == self.get_input_layer(
                    input_index).depth
2781 2782 2783 2784 2785

        self.set_layer_size(self.get_input_layer(0).size)
        self.set_layer_height_width(self.get_input_layer(0).height, \
                                        self.get_input_layer(0).width)
        self.set_layer_depth(self.get_input_layer(0).depth)
Z
zhangjinchao01 已提交
2786 2787
        self.create_bias_parameter(bias, self.config.size)

Q
qijun 已提交
2788

Z
zhangjinchao01 已提交
2789 2790
@config_layer('agent')
class AgentLayer(LayerBase):
Q
qijun 已提交
2791 2792 2793 2794
    def __init__(self, name, size, device=None):
        super(AgentLayer, self).__init__(
            name, 'agent', size, inputs=[], device=device)

Z
zhangjinchao01 已提交
2795 2796 2797

@config_layer('gather_agent')
class GatherAgentLayer(LayerBase):
Q
qijun 已提交
2798
    def __init__(self, name, size, device=None):
Z
zhangjinchao01 已提交
2799 2800 2801
        super(GatherAgentLayer, self).__init__(
            name, 'gather_agent', size, inputs=[], device=device)

Q
qijun 已提交
2802

Z
zhangjinchao01 已提交
2803 2804
@config_layer('scatter_agent')
class ScatterAgentLayer(LayerBase):
2805
    def __init__(self, name, size, width=None, height=None, device=None):
Z
zhangjinchao01 已提交
2806 2807
        super(ScatterAgentLayer, self).__init__(
            name, 'scatter_agent', size, inputs=[], device=device)
2808 2809
        if height and width:
            self.set_layer_height_width(height, width)
Z
zhangjinchao01 已提交
2810

Q
qijun 已提交
2811

Z
zhangjinchao01 已提交
2812 2813
@config_layer('multiplex')
class MultiplexLayer(LayerBase):
Q
qijun 已提交
2814 2815 2816 2817 2818
    def __init__(self, name, inputs, size, device=None):
        super(MultiplexLayer, self).__init__(
            name, 'multiplex', size, inputs=inputs, device=device)
        config_assert(
            len(inputs) > 2, 'MultiplexLayer should have more than 2 inputs.')
Z
zhangjinchao01 已提交
2819
        for i in range(1, len(inputs)):
Q
qijun 已提交
2820 2821 2822 2823 2824
            config_assert(
                self.get_input_layer(i).size == size,
                "All the input layers except the first one should"
                "have the same size as the MultiplexLayer.")

Z
zhangjinchao01 已提交
2825 2826

@config_func
2827 2828 2829 2830
def Link(name, has_subseq=False):
    """
    Still keeping has_subseq for backward compatibility
    """
Z
zhangjinchao01 已提交
2831 2832 2833 2834
    link_config = LinkConfig()
    link_config.link_name = name
    return link_config

Q
qijun 已提交
2835

Z
zhangjinchao01 已提交
2836 2837
# memory for recurrent layer group.
# *name* and *size* are actual layer's name and size.
2838 2839 2840 2841
# If *name* is None, need to provide *memory_name* and need to use
# SetMemoryInput() later to specify the layer which this memory remembers.
#
# return the name of the memory,
Z
zhangjinchao01 已提交
2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852
# use this name if you assign the memory as other layer's input
#
# boot frame of memory is zeroed by default,
# or initialize by boot layer output if *boot_layer* set,
# or initialize by trainable bias if *boot_bias* set,
# or initialize by a constant id if *boot_with_const_id* set
#
# Memory can be a sequence if *is_sequence* set, this type of memory
# can only be initailized by a *boot_layer* which is a sequence.
#
@config_func
2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864
def Memory(name,
           size,
           is_sequence=False,
           boot_layer=None,
           boot_bias=False,
           boot_bias_active_type="",
           boot_with_const_id=None,
           memory_name=None):
    if not memory_name:
        config_assert(name is not None, "name needs cannot be None")
        memory_name = name + "+delay1"
    agent_name = memory_name
2865
    agent_layer = AgentLayer(agent_name, size)
Z
zhangjinchao01 已提交
2866
    config_assert(g_current_submodel.is_recurrent_layer_group,
Q
qijun 已提交
2867
                  'Memory should be used in recurrent layer group only')
Z
zhangjinchao01 已提交
2868
    memory = g_current_submodel.memories.add()
2869 2870
    if name is not None:
        memory.layer_name = MakeLayerNameInSubmodel(name)
Z
zhangjinchao01 已提交
2871
    memory.link_name = MakeLayerNameInSubmodel(agent_name)
Q
qijun 已提交
2872
    options = sum((boot_layer is not None, bool(boot_bias),
Z
zhangjinchao01 已提交
2873
                   boot_with_const_id is not None))
Q
qijun 已提交
2874 2875 2876 2877
    config_assert(
        options <= 1,
        'take one option at most from boot_layer, boot_bias, or boot_with_const_id'
    )
Z
zhangjinchao01 已提交
2878 2879 2880
    if boot_layer is not None:
        boot_layer = MakeLayerNameInParentSubmodel(boot_layer)
        config_assert(boot_layer in g_layer_map,
Q
qijun 已提交
2881 2882
                      'boot_layer "%s" does not correspond to a layer name' %
                      boot_layer)
Z
zhangjinchao01 已提交
2883 2884 2885
        memory.boot_layer_name = boot_layer
    elif boot_bias:
        memory.boot_bias_parameter_name = agent_layer.create_bias_parameter(
Q
qijun 已提交
2886
            boot_bias, size, for_self=False)
Z
zhangjinchao01 已提交
2887 2888 2889 2890 2891
        memory.boot_bias_active_type = boot_bias_active_type
    elif boot_with_const_id is not None:
        memory.boot_with_const_id = boot_with_const_id
    return agent_name

Q
qijun 已提交
2892

2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903
@config_func
def SetMemoryInput(memory_name, layer_name):
    memory_name = MakeLayerNameInSubmodel(memory_name)
    layer_name = MakeLayerNameInSubmodel(layer_name)
    for mem in g_current_submodel.memories:
        if mem.link_name == memory_name:
            mem.layer_name = layer_name
            return
    logger.fatal("Nonexistent memory name: " + memory_name)


Z
zhangjinchao01 已提交
2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914
# Generator for recurrent layer group, to use it:
#  1. define a id layer as output of layer group
#  2. define a memory of this id layer, and assign a boot id(begin of sequence)
#  3. define a eos check layer and fill its name in generator's *eos_layer_name*
# Sequence generation will stop when eos check return 1 or *max_num_frames* reached.
# If *beam_size* is greater than one, generator will use beam search.
#   in beam search, if *num_results_per_sample* set, one sample sequence can output
#   multiple results each with a probility.
@config_func
def Generator(
        max_num_frames,
Q
qijun 已提交
2915 2916 2917 2918
        eos_layer_name="eos_check",
        num_results_per_sample=1,
        beam_size=1,
        log_prob=None, ):
Z
zhangjinchao01 已提交
2919 2920 2921 2922 2923 2924 2925 2926 2927
    generator_config = GeneratorConfig()
    generator_config.max_num_frames = max_num_frames
    generator_config.eos_layer_name = eos_layer_name
    generator_config.num_results_per_sample = num_results_per_sample
    generator_config.beam_size = beam_size
    if log_prob is not None:
        generator_config.log_prob = log_prob
    return generator_config

Q
qijun 已提交
2928

Z
zhangjinchao01 已提交
2929 2930
@config_layer('expand')
class ExpandLayer(LayerBase):
2931
    def __init__(self, name, inputs, trans_type='non-seq', bias=False, **xargs):
Q
qijun 已提交
2932
        super(ExpandLayer, self).__init__(
2933
            name, 'expand', 0, inputs=inputs, **xargs)
Q
qijun 已提交
2934 2935 2936 2937 2938 2939 2940 2941
        config_assert(
            len(self.inputs) == 2, 'ExpandLayer takes 2 and only 2 inputs')
        self.config.trans_type = trans_type
        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
        self.set_layer_size(self.get_input_layer(0).size)
        self.create_bias_parameter(bias, self.config.size)

Z
zhangjinchao01 已提交
2942 2943 2944

@config_layer('featmap_expand')
class FeatMapExpandLayer(LayerBase):
X
xuwei06 已提交
2945 2946 2947 2948 2949
    def __init__(self,
                 name,
                 inputs,
                 num_filters=None,
                 as_row_vector=True,
X
xuwei06 已提交
2950 2951
                 bias=False,
                 **xargs):
Q
qijun 已提交
2952
        super(FeatMapExpandLayer, self).__init__(
X
xuwei06 已提交
2953
            name, 'featmap_expand', 0, inputs=inputs, **xargs)
Q
qijun 已提交
2954 2955 2956
        config_assert(
            len(self.inputs) == 1, 'ExpandLayer takes 1 and only 1 inputs')
        if num_filters is not None:
Z
zhangjinchao01 已提交
2957
            self.config.num_filters = num_filters
Q
qijun 已提交
2958
        else:
Z
zhangjinchao01 已提交
2959
            logger.fatal("FeatMapExpandLayer must specify num_filters.")
X
xuwei06 已提交
2960 2961
        if not as_row_vector:
            self.config.user_arg = "as_col_vec"
Q
qijun 已提交
2962
        self.set_layer_size(self.get_input_layer(0).size * num_filters)
Z
zhangjinchao01 已提交
2963 2964 2965 2966


@config_layer('max')
class MaxLayer(LayerBase):
Q
qijun 已提交
2967 2968 2969 2970 2971
    def __init__(self,
                 name,
                 inputs,
                 trans_type='non-seq',
                 bias=False,
2972
                 output_max_index=None,
2973
                 stride=-1,
2974
                 **xargs):
2975
        super(MaxLayer, self).__init__(name, 'max', 0, inputs=inputs, **xargs)
Z
zhangjinchao01 已提交
2976
        config_assert(len(self.inputs) == 1, 'MaxLayer must have 1 input')
2977 2978
        if trans_type == 'seq':
            config_assert(stride == -1, 'subseq does not support stride window')
Q
qijun 已提交
2979
        self.config.trans_type = trans_type
2980
        self.config.seq_pool_stride = stride
Z
zhangjinchao01 已提交
2981 2982 2983 2984
        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
            self.set_layer_size(input_layer.size)
        self.create_bias_parameter(bias, self.config.size)
2985 2986
        if output_max_index is not None:
            self.config.output_max_index = output_max_index
Z
zhangjinchao01 已提交
2987 2988 2989 2990


@config_layer('maxid')
class MaxIdLayer(LayerBase):
Q
qijun 已提交
2991
    def __init__(self, name, inputs, beam_size=None, device=None):
Z
zhangjinchao01 已提交
2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008
        super(MaxIdLayer, self).__init__(
            name, 'maxid', 0, inputs=inputs, device=device)
        config_assert(len(self.inputs) == 1, 'MaxIdLayer must have 1 input')
        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
            self.set_layer_size(input_layer.size)

        if beam_size is None:
            global g_current_submodel
            if g_current_submodel.HasField("generator"):
                self.config.beam_size = g_current_submodel.generator.beam_size
        else:
            self.config.beam_size = beam_size


@config_layer('eos_id')
class EosIdLayer(LayerBase):
Q
qijun 已提交
3009
    def __init__(self, name, inputs, eos_id, device=None):
Z
zhangjinchao01 已提交
3010 3011 3012
        super(EosIdLayer, self).__init__(
            name, 'eos_id', 0, inputs=inputs, device=device)
        config_assert(len(self.inputs) == 1, 'EosIdLayer must have 1 input')
Q
qijun 已提交
3013
        self.set_layer_size(2)  # boolean output
Z
zhangjinchao01 已提交
3014 3015
        self.config.eos_id = eos_id

Q
qijun 已提交
3016

Z
zhangjinchao01 已提交
3017 3018
@config_layer('seqlastins')
class SequenceLastInstanceLayer(LayerBase):
Q
qijun 已提交
3019 3020 3021 3022
    def __init__(self,
                 name,
                 inputs,
                 trans_type='non-seq',
3023
                 bias=False,
3024
                 stride=-1,
3025
                 **xargs):
Q
qijun 已提交
3026
        super(SequenceLastInstanceLayer, self).__init__(
X
xuwei06 已提交
3027
            name, 'seqlastins', 0, inputs=inputs, **xargs)
Q
qijun 已提交
3028 3029
        config_assert(
            len(inputs) == 1, 'SequenceLastInstanceLayer must have 1 input')
3030
        if trans_type == 'seq':
L
Luo Tao 已提交
3031
            config_assert(stride == -1, 'subseq does not support stride window')
Q
qijun 已提交
3032
        self.config.trans_type = trans_type
3033 3034
        self.config.seq_pool_stride = stride
        self.set_layer_size(self.get_input_layer(0).size)
Z
zhangjinchao01 已提交
3035 3036
        self.create_bias_parameter(bias, self.config.size)

Q
qijun 已提交
3037

Z
zhangjinchao01 已提交
3038 3039
@config_layer('seqfirstins')
class SequenceFirstInstanceLayer(SequenceLastInstanceLayer):
3040 3041 3042 3043 3044
    def __init__(self,
                 name,
                 inputs,
                 trans_type='non-seq',
                 bias=False,
3045
                 stride=-1,
3046
                 **xargs):
Q
qijun 已提交
3047
        super(SequenceFirstInstanceLayer, self).__init__(
3048 3049 3050 3051 3052 3053
            name,
            inputs=inputs,
            trans_type=trans_type,
            bias=bias,
            stride=stride,
            **xargs)
Z
zhangjinchao01 已提交
3054 3055
        self.config.select_first = True

Q
qijun 已提交
3056

Z
zhangjinchao01 已提交
3057 3058
@config_layer('seqconcat')
class SequenceConcatLayer(LayerBase):
X
xuwei06 已提交
3059
    def __init__(self, name, inputs, bias=False, **xargs):
Q
qijun 已提交
3060
        super(SequenceConcatLayer, self).__init__(
X
xuwei06 已提交
3061
            name, 'seqconcat', 0, inputs=inputs, **xargs)
Q
qijun 已提交
3062 3063
        config_assert(
            len(inputs) == 2, 'SequenceConcatLayer must have 2 inputs')
Z
zhangjinchao01 已提交
3064 3065 3066 3067 3068
        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
            self.set_layer_size(input_layer.size)
        self.create_bias_parameter(bias, self.config.size)

Q
qijun 已提交
3069

Z
zhangjinchao01 已提交
3070 3071
@config_layer('seqreshape')
class SequenceReshapeLayer(LayerBase):
X
xuwei06 已提交
3072
    def __init__(self, name, size, inputs, bias=False, **xargs):
Q
qijun 已提交
3073
        super(SequenceReshapeLayer, self).__init__(
X
xuwei06 已提交
3074
            name, 'seqreshape', size, inputs=inputs, **xargs)
Q
qijun 已提交
3075 3076
        config_assert(
            len(inputs) == 1, 'SequenceReshapeLayer must have 1 inputs')
Z
zhangjinchao01 已提交
3077 3078 3079
        self.set_layer_size(size)
        self.create_bias_parameter(bias, size)

Q
qijun 已提交
3080

Z
zhangjinchao01 已提交
3081 3082
@config_layer('subseq')
class SubSequenceLayer(LayerBase):
X
xuwei06 已提交
3083
    def __init__(self, name, inputs, bias=False, **xargs):
Q
qijun 已提交
3084
        super(SubSequenceLayer, self).__init__(
X
xuwei06 已提交
3085
            name, 'subseq', 0, inputs=inputs, **xargs)
Z
zhangjinchao01 已提交
3086 3087 3088 3089 3090 3091
        config_assert(len(inputs) == 3, 'SubSequenceLayer must have 3 inputs')
        input_layer0 = self.get_input_layer(0)
        size = input_layer0.size
        self.set_layer_size(size)
        self.create_bias_parameter(bias, size)

Q
qijun 已提交
3092

3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121
@config_layer('seq_slice')
class SeqSliceLayer(LayerBase):
    def __init__(self, name, inputs, starts, ends, bias=False, **xargs):
        if isinstance(inputs, list):
            assert len(inputs) == 1, ('the first input of sequence slice layer '
                                      'is a single sequence input.')
        else:
            inputs = [inputs]

        if starts is not None:
            if isinstance(starts, list):
                assert len(starts) == 1, (
                    'the start indices for sequence slice layer cannot '
                    'be a list having more than one element.')
                starts = starts[0]
            inputs.append(starts)

        if ends is not None:
            if isinstance(ends, list):
                assert len(ends) == 1, (
                    'the end indices for sequence slice layer cannot '
                    'be a list having more than one element.')
                ends = ends[0]
            inputs.append(ends)
        assert len(inputs) >= 2, (
            'the sequence slice layer has at least two inputs.')

        super(SeqSliceLayer, self).__init__(
            name, 'seq_slice', 0, inputs=inputs, **xargs)
3122

3123 3124 3125 3126 3127 3128 3129 3130 3131 3132
        input_layer0 = self.get_input_layer(0)
        size = input_layer0.size
        self.set_layer_size(size)

        if len(inputs) == 3:
            assert (
                self.get_input_layer(1).size == self.get_input_layer(2).size), (
                    'If start and end indices are both given to'
                    'sequence slice layer, they should have the same width.')
        elif len(inputs) == 2:
C
caoying03 已提交
3133
            self.config.select_first = (starts is not None)
3134 3135


C
caoying03 已提交
3136 3137
@config_layer('sub_nested_seq')
class SubNestedSequenceLayer(LayerBase):
3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149
    def __init__(self, name, inputs, selected_indices, bias=False, **xargs):
        if isinstance(inputs, list):
            assert len(inputs) == 1, ('the first input of sub_nested_seq '
                                      'layer is a single nested sequence.')
            inputs = inputs[0]
        if isinstance(selected_indices, list):
            assert len(selected_indices) == 1, (
                'the second input of '
                'sub_nested_seq layer is a single layer which is a '
                'set of selected indices.')
            selected_indices = selected_indices[0]

C
caoying03 已提交
3150
        super(SubNestedSequenceLayer, self).__init__(
3151 3152 3153 3154 3155
            name,
            'sub_nested_seq',
            0,
            inputs=[inputs, selected_indices],
            **xargs)
C
caoying03 已提交
3156 3157 3158 3159 3160
        input_layer0 = self.get_input_layer(0)
        size = input_layer0.size
        self.set_layer_size(size)


Z
zhangjinchao01 已提交
3161 3162
@config_layer('out_prod')
class OuterProdLayer(LayerBase):
Q
qijun 已提交
3163 3164 3165
    def __init__(self, name, inputs, device=None):
        super(OuterProdLayer, self).__init__(
            name, 'out_prod', 0, inputs=inputs, device=device)
Z
zhangjinchao01 已提交
3166 3167 3168 3169 3170
        config_assert(len(inputs) == 2, 'OuterProdLayer must have 2 inputs')
        input_layer0 = self.get_input_layer(0)
        input_layer1 = self.get_input_layer(1)
        self.set_layer_size(input_layer0.size * input_layer1.size)

Q
qijun 已提交
3171

Z
zhangjinchao01 已提交
3172 3173
@config_layer('power')
class PowerLayer(LayerBase):
Q
qijun 已提交
3174 3175 3176
    def __init__(self, name, inputs, device=None):
        super(PowerLayer, self).__init__(
            name, 'power', 0, inputs=inputs, device=device)
Z
zhangjinchao01 已提交
3177 3178 3179 3180
        config_assert(len(inputs) == 2, 'PowerLayer must have 2 inputs')
        input_layer1 = self.get_input_layer(1)
        self.set_layer_size(input_layer1.size)
        input_layer0 = self.get_input_layer(0)
Q
qijun 已提交
3181 3182 3183
        config_assert(1 == input_layer0.size,
                      'The left input is the exponent and should be of size 1')

Z
zhangjinchao01 已提交
3184 3185 3186

@config_layer('slope_intercept')
class SlopeInterceptLayer(LayerBase):
Q
qijun 已提交
3187 3188 3189
    def __init__(self, name, inputs, slope=1.0, intercept=0.0, device=None):
        super(SlopeInterceptLayer, self).__init__(
            name, 'slope_intercept', 0, inputs=inputs, device=device)
Z
zhangjinchao01 已提交
3190 3191 3192 3193 3194 3195
        self.config.slope = slope
        self.config.intercept = intercept
        config_assert(len(inputs) == 1, 'SlopeInterceptLayer must have 1 input')
        input_layer0 = self.get_input_layer(0)
        self.set_layer_size(input_layer0.size)

Q
qijun 已提交
3196

Z
zhangjinchao01 已提交
3197 3198
@config_layer('scaling')
class ScalingLayer(LayerBase):
Q
qijun 已提交
3199 3200 3201
    def __init__(self, name, inputs, device=None):
        super(ScalingLayer, self).__init__(
            name, 'scaling', 0, inputs=inputs, device=device)
Z
zhangjinchao01 已提交
3202 3203 3204 3205
        config_assert(len(inputs) == 2, 'ScalingLayer must have 2 inputs')
        input_layer1 = self.get_input_layer(1)
        self.set_layer_size(input_layer1.size)
        input_layer0 = self.get_input_layer(0)
Q
qijun 已提交
3206 3207 3208
        config_assert(1 == input_layer0.size,
                      'The left input should be of size 1')

Z
zhangjinchao01 已提交
3209 3210 3211

@config_layer('conv_shift')
class ConvShiftLayer(LayerBase):
Q
qijun 已提交
3212 3213 3214
    def __init__(self, name, inputs, device=None):
        super(ConvShiftLayer, self).__init__(
            name, 'conv_shift', 0, inputs=inputs, device=device)
Z
zhangjinchao01 已提交
3215 3216 3217 3218
        config_assert(len(inputs) == 2, 'ConvShiftLayer must have 2 inputs')
        input_layer0 = self.get_input_layer(0)
        self.set_layer_size(input_layer0.size)

Q
qijun 已提交
3219

Z
zhangjinchao01 已提交
3220 3221
@config_layer('convex_comb')
class ConvexCombinationLayer(LayerBase):
Q
qijun 已提交
3222
    def __init__(self, name, size, inputs, device=None):
Z
zhangjinchao01 已提交
3223
        super(ConvexCombinationLayer, self).__init__(
Q
qijun 已提交
3224 3225 3226
            name, 'convex_comb', size, inputs=inputs, device=device)
        config_assert(
            len(self.inputs) == 2, 'ConvexCombinationLayer must have 2 inputs')
3227 3228 3229
        config_assert(
            size * self.get_input_layer(0).size == self.get_input_layer(1).size,
            'Wrong input size for ConvexCombinationLayer')
Z
zhangjinchao01 已提交
3230 3231
        self.set_layer_size(size)

Q
qijun 已提交
3232

Z
zhangjinchao01 已提交
3233 3234
@config_layer('interpolation')
class InterpolationLayer(LayerBase):
Q
qijun 已提交
3235
    def __init__(self, name, inputs, device=None):
Z
zhangjinchao01 已提交
3236 3237
        super(InterpolationLayer, self).__init__(
            name, 'interpolation', 0, inputs=inputs, device=device)
Q
qijun 已提交
3238 3239
        config_assert(
            len(self.inputs) == 3, 'InterpolationLayer must have 3 inputs')
Z
zhangjinchao01 已提交
3240 3241 3242 3243 3244 3245 3246 3247
        input_layer0 = self.get_input_layer(0)
        input_layer1 = self.get_input_layer(1)
        input_layer2 = self.get_input_layer(2)
        self.set_layer_size(input_layer1.size)
        config_assert(input_layer0.size == 1, 'weight should be of size 1')
        config_assert(input_layer1.size == input_layer2.size,
                      'the two vector inputs should be of the same size')

Q
qijun 已提交
3248

L
liaogang 已提交
3249 3250
@config_layer('bilinear_interp')
class BilinearInterpLayer(LayerBase):
Q
qijun 已提交
3251
    def __init__(self, name, inputs, **xargs):
L
liaogang 已提交
3252
        super(BilinearInterpLayer, self).__init__(
L
liaogang 已提交
3253
            name, 'bilinear_interp', 0, inputs=inputs, **xargs)
L
liaogang 已提交
3254
        input_layer = self.get_input_layer(0)
L
Luo Tao 已提交
3255 3256 3257 3258
        conf = self.config.inputs[0].bilinear_interp_conf
        parse_bilinear(self.inputs[0].bilinear_interp, input_layer.name, conf)
        self.set_cnn_layer(name, conf.out_size_y, conf.out_size_x,
                           conf.image_conf.channels)
Q
qijun 已提交
3259

L
liaogang 已提交
3260

Z
zhangjinchao01 已提交
3261 3262
@config_layer('sum_to_one_norm')
class SumToOneNormLayer(LayerBase):
Q
qijun 已提交
3263
    def __init__(self, name, inputs, device=None):
Z
zhangjinchao01 已提交
3264
        super(SumToOneNormLayer, self).__init__(
Q
qijun 已提交
3265 3266 3267
            name, 'sum_to_one_norm', 0, inputs=inputs, device=device)
        config_assert(
            len(self.inputs) == 1, 'SumToOneNormLayer must have 1 input')
Z
zhangjinchao01 已提交
3268 3269 3270
        input_layer0 = self.get_input_layer(0)
        self.set_layer_size(input_layer0.size)

Q
qijun 已提交
3271

G
guosheng 已提交
3272 3273
@config_layer('row_l2_norm')
class RowL2NormLayer(LayerBase):
3274
    def __init__(self, name, inputs, **xargs):
G
guosheng 已提交
3275
        super(RowL2NormLayer, self).__init__(
3276
            name, 'row_l2_norm', 0, inputs=inputs, **xargs)
G
guosheng 已提交
3277
        config_assert(len(self.inputs) == 1, 'RowL2NormLayer must have 1 input')
3278 3279
        input_layer = self.get_input_layer(0)
        self.set_layer_size(input_layer.size)
G
guosheng 已提交
3280 3281


Z
zhangjinchao01 已提交
3282 3283
@config_layer('cos_vm')
class CosSimVecMatLayer(LayerBase):
Q
qijun 已提交
3284
    def __init__(self, name, size, inputs, cos_scale=1.0, device=None):
Z
zhangjinchao01 已提交
3285
        super(CosSimVecMatLayer, self).__init__(
Q
qijun 已提交
3286
            name, 'cos_vm', size, inputs=inputs, device=device)
Z
zhangjinchao01 已提交
3287
        self.config.cos_scale = cos_scale
Q
qijun 已提交
3288 3289
        config_assert(
            len(self.inputs) == 2, 'CosSimVecMatLayer must have 2 inputs')
3290 3291 3292
        config_assert(
            size * self.get_input_layer(0).size == self.get_input_layer(1).size,
            'Wrong input size for CosSimVecMatLayer')
Z
zhangjinchao01 已提交
3293

Q
qijun 已提交
3294

Z
zhangjinchao01 已提交
3295 3296
@config_layer('sampling_id')
class SamplingIdLayer(LayerBase):
Q
qijun 已提交
3297
    def __init__(self, name, inputs, device=None):
Z
zhangjinchao01 已提交
3298 3299
        super(SamplingIdLayer, self).__init__(
            name, 'sampling_id', 0, inputs=inputs, device=device)
Q
qijun 已提交
3300 3301
        config_assert(
            len(self.inputs) == 1, 'SamplingIdLayer must have 1 input')
Z
zhangjinchao01 已提交
3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313
        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
            self.set_layer_size(input_layer.size)


# AverageLayer: "average" for each sample within a sequence.
# average_stratrgy: set to one of the following:
# 'average': plain average.
# 'sum': sum each sample instead of average (which is divide by sample_num).
# 'squarerootn': sum each sample, but divide by sqrt(sample_num).
@config_layer('average')
class AverageLayer(LayerBase):
Q
qijun 已提交
3314 3315 3316 3317 3318
    def __init__(self,
                 name,
                 inputs,
                 average_strategy='average',
                 trans_type='non-seq',
3319
                 bias=False,
3320
                 stride=-1,
3321
                 **xargs):
Q
qijun 已提交
3322
        super(AverageLayer, self).__init__(
X
xuwei06 已提交
3323
            name, 'average', 0, inputs=inputs, **xargs)
Z
zhangjinchao01 已提交
3324
        self.config.average_strategy = average_strategy
3325 3326
        if trans_type == 'seq':
            config_assert(stride == -1, 'subseq does not support stride window')
Q
qijun 已提交
3327
        self.config.trans_type = trans_type
3328
        self.config.seq_pool_stride = stride
Z
zhangjinchao01 已提交
3329 3330 3331 3332 3333 3334
        config_assert(len(inputs) == 1, 'AverageLayer must have 1 input')
        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
            self.set_layer_size(input_layer.size)
        self.create_bias_parameter(bias, self.config.size)

Q
qijun 已提交
3335

Z
zhangjinchao01 已提交
3336 3337
@config_layer('cos')
class CosSimLayer(LayerBase):
3338
    def __init__(self, name, inputs, cos_scale=1, device=None):
Z
zhangjinchao01 已提交
3339 3340 3341 3342 3343 3344
        super(CosSimLayer, self).__init__(
            name, 'cos', 1, inputs=inputs, device=device)
        config_assert(len(self.inputs) == 2, 'CosSimLayer must have 2 inputs')
        config_assert(
            self.get_input_layer(0).size == self.get_input_layer(1).size,
            'inputs of CosSimLayer must have same dim')
3345
        self.config.cos_scale = cos_scale
Z
zhangjinchao01 已提交
3346 3347 3348 3349


@config_layer('tensor')
class TensorLayer(LayerBase):
3350
    def __init__(self, name, size, inputs, bias=True, **xargs):
Q
qijun 已提交
3351
        super(TensorLayer, self).__init__(
3352
            name, 'tensor', size, inputs=inputs, **xargs)
Z
zhangjinchao01 已提交
3353 3354
        config_assert(len(self.inputs) == 2, 'TensorLayer must have 2 inputs')
        config_assert(size > 0, 'size must be positive')
Q
qijun 已提交
3355 3356
        config_assert(inputs[1].parameter_name == None,
                      'second parameter should be None.')
Z
zhangjinchao01 已提交
3357 3358 3359 3360 3361 3362 3363 3364 3365 3366
        input_layer0 = self.get_input_layer(0)
        input_layer1 = self.get_input_layer(1)
        psize = size * input_layer0.size * input_layer1.size
        dims = [input_layer0.size, input_layer1.size, size]
        self.create_input_parameter(0, psize, dims)
        self.create_bias_parameter(bias, size)


@config_layer('mixed')
class MixedLayer(LayerBase):
C
caoying03 已提交
3367
    def __init__(self, name, inputs, size=0, bias=True, **xargs):
Z
zhangjinchao01 已提交
3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384
        config_assert(inputs, 'inputs cannot be empty')
        super(MixedLayer, self).__init__(
            name, 'mixed', size, inputs=inputs, **xargs)
        operator_input_index = []
        for operator in self.operators:
            operator_conf = operator.operator_conf
            for i in xrange(1, len(operator.input_layer_names)):
                input_index = len(self.config.inputs)
                operator_conf.input_indices.append(input_index)
                input_config = Input(operator.input_layer_names[i])
                self.inputs.append(input_config)
                layer_input = self.config.inputs.add()
                layer_input.input_layer_name = input_config.input_layer_name
            for input_index in operator_conf.input_indices:
                input_layer = self.get_input_layer(input_index)
                operator_conf.input_sizes.append(input_layer.size)
                operator_input_index.append(input_index)
3385
            if self.config.size == 0:
Z
zhangjinchao01 已提交
3386 3387 3388
                size = operator.calc_output_size(operator_conf.input_sizes)
                if size != 0:
                    self.set_layer_size(size)
3389
            else:
3390 3391
                sz = operator.calc_output_size(operator_conf.input_sizes)
                if sz != 0:
Q
qijun 已提交
3392 3393 3394 3395
                    config_assert(
                        sz == self.config.size,
                        "different inputs have different size: %s vs. %s" %
                        (sz, self.config.size))
Z
zhangjinchao01 已提交
3396 3397 3398 3399
        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
            input = self.inputs[input_index]
            if input_index not in operator_input_index:
Q
qijun 已提交
3400 3401 3402
                config_assert(
                    isinstance(input, Projection),
                    "input should be projection or operation")
3403
            if self.config.size == 0 and isinstance(input, Projection):
Z
zhangjinchao01 已提交
3404 3405 3406
                size = input.calc_output_size(input_layer)
                if size != 0:
                    self.set_layer_size(size)
3407
            elif isinstance(input, Projection):
Q
qijun 已提交
3408 3409 3410 3411 3412 3413
                sz = input.calc_output_size(input_layer)
                if sz != 0:
                    config_assert(
                        sz == self.config.size,
                        "different inputs have different size: %s vs. %s" %
                        (sz, self.config.size))
Z
zhangjinchao01 已提交
3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424
        config_assert(size != 0, "size is not set")

        for input_index in xrange(len(self.inputs)):
            input = self.inputs[input_index]
            if isinstance(input, Projection):
                input_layer = self.get_input_layer(input_index)
                input.proj_conf.input_size = input_layer.size
                input.proj_conf.output_size = size

                input_config = self.config.inputs[input_index]
                input_config.proj_conf.CopyFrom(input.proj_conf)
Q
qijun 已提交
3425 3426
                input_config.proj_conf.name = gen_parameter_name(name,
                                                                 input_index)
Z
zhangjinchao01 已提交
3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437
                psize = input.calc_parameter_size(input_layer.size, size)
                dims = input.calc_parameter_dims(input_layer.size, size)
                self.create_input_parameter(input_index, psize, dims)

        for operator in self.operators:
            operator_conf = operator.operator_conf
            operator_conf.output_size = self.config.size
            operator.check_dims()
            record_operator_conf = self.config.operator_confs.add()
            record_operator_conf.CopyFrom(operator_conf)

3438 3439 3440 3441 3442 3443
        psize = self.config.size
        if isinstance(self.inputs[0], ConvProjection):
            self.config.shared_biases = True
            psize = 0
            for input in self.inputs:
                psize += input.calc_bias_size()
Z
zhangjinchao01 已提交
3444

3445 3446 3447
        if bias:
            self.config.bias_size = psize
            self.create_bias_parameter(bias, psize)
Z
zhangjinchao01 已提交
3448

Q
qijun 已提交
3449

Z
zhangjinchao01 已提交
3450 3451
# like MixedLayer, but no bias parameter
@config_func
Q
qijun 已提交
3452
def ExpressionLayer(name, inputs, **xargs):
Z
zhangjinchao01 已提交
3453 3454
    MixedLayer(name, inputs, bias=False, **xargs)

Q
qijun 已提交
3455

Z
zhangjinchao01 已提交
3456 3457
@config_layer('concat')
class ConcatenateLayer(LayerBase):
Q
qijun 已提交
3458
    def __init__(self, name, inputs, bias=False, **xargs):
Z
zhangjinchao01 已提交
3459
        config_assert(inputs, 'inputs cannot be empty')
3460
        config_assert(not bias, 'ConcatenateLayer cannot support bias.')
Z
zhangjinchao01 已提交
3461 3462 3463 3464
        super(ConcatenateLayer, self).__init__(
            name, 'concat', 0, inputs=inputs, **xargs)
        size = 0
        for input_index in xrange(len(self.inputs)):
3465 3466 3467 3468 3469 3470
            assert self.get_input_layer(0).height == self.get_input_layer(
                input_index).height
            assert self.get_input_layer(0).width == self.get_input_layer(
                input_index).width
            assert self.get_input_layer(0).depth == self.get_input_layer(
                input_index).depth
Z
zhangjinchao01 已提交
3471 3472
            input_layer = self.get_input_layer(input_index)
            input = self.inputs[input_index]
Q
qijun 已提交
3473
            if self.config.size == 0:
Z
zhangjinchao01 已提交
3474 3475
                size += input_layer.size

3476 3477 3478
        self.set_layer_height_width(self.get_input_layer(0).height, \
                                    self.get_input_layer(0).width)
        self.set_layer_depth(self.get_input_layer(0).depth)
Z
zhangjinchao01 已提交
3479 3480
        self.set_layer_size(size)

Q
qijun 已提交
3481

Z
zhangjinchao01 已提交
3482 3483 3484
# like concat layer, but each input layer was processed by a Projection.
@config_layer('concat2')
class ConcatenateLayer2(LayerBase):
Q
qijun 已提交
3485
    def __init__(self, name, inputs, bias=False, **xargs):
Z
zhangjinchao01 已提交
3486 3487 3488
        config_assert(inputs, 'inputs cannot be empty')
        super(ConcatenateLayer2, self).__init__(
            name, 'concat2', 0, inputs=inputs, **xargs)
3489 3490

        if isinstance(self.inputs[0], ConvProjection):
Q
qijun 已提交
3491 3492 3493 3494 3495 3496
            for input_index in xrange(len(self.inputs) - 1):
                input = self.inputs[input_index + 1]
                config_assert(
                    isinstance(input, ConvProjection),
                    "The first input of ConcatenateLayer2 is ConvProjection, "
                    "the other inputs should also be ConvProjection.")
3497

Z
zhangjinchao01 已提交
3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517
        size = 0
        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
            input = self.inputs[input_index]
            output_size = input.calc_output_size(input_layer)
            config_assert(output_size != 0, "proj output size is not set")
            size += output_size

        self.set_layer_size(size)

        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
            input = self.inputs[input_index]
            input.proj_conf.input_size = input_layer.size
            input.proj_conf.output_size = input.calc_output_size(input_layer)

            input_config = self.config.inputs[input_index]
            input_config.proj_conf.CopyFrom(input.proj_conf)
            input_config.proj_conf.name = gen_parameter_name(name, input_index)
            psize = input.calc_parameter_size(input.proj_conf.input_size,
Q
qijun 已提交
3518
                                              input.proj_conf.output_size)
Z
zhangjinchao01 已提交
3519
            dims = input.calc_parameter_dims(input.proj_conf.input_size,
Q
qijun 已提交
3520
                                             input.proj_conf.output_size)
Z
zhangjinchao01 已提交
3521 3522
            self.create_input_parameter(input_index, psize, dims)

3523 3524 3525 3526 3527 3528 3529
        psize = self.config.size
        if isinstance(self.inputs[0], ConvProjection):
            self.config.shared_biases = True
            psize = 0
            for input in self.inputs:
                psize += input.calc_bias_size()

3530 3531 3532
        if bias:
            self.config.bias_size = psize
            self.create_bias_parameter(bias, psize)
3533

Q
qijun 已提交
3534

Z
zhangjinchao01 已提交
3535 3536
@config_layer('recurrent')
class RecurrentLayer(LayerBase):
Q
qijun 已提交
3537
    def __init__(self, name, inputs, reversed=False, bias=True, **xargs):
Y
Yu Yang 已提交
3538 3539
        super(RecurrentLayer, self).__init__(name, 'recurrent', 0, inputs,
                                             **xargs)
Z
zhangjinchao01 已提交
3540 3541 3542 3543 3544 3545 3546 3547 3548
        config_assert(len(self.inputs) == 1, 'RecurrentLayer must have 1 input')
        input_layer = self.get_input_layer(0)
        size = input_layer.size
        self.set_layer_size(size)
        self.config.reversed = reversed
        dims = [size, size]
        self.create_input_parameter(0, size * size, dims)
        self.create_bias_parameter(bias, self.config.size)

Q
qijun 已提交
3549

Z
zhangjinchao01 已提交
3550 3551
@config_layer('lstmemory')
class LstmLayer(LayerBase):
Q
qijun 已提交
3552 3553 3554 3555 3556 3557 3558 3559
    def __init__(self,
                 name,
                 inputs,
                 reversed=False,
                 active_gate_type="sigmoid",
                 active_state_type="sigmoid",
                 bias=True,
                 **xargs):
Z
zhangjinchao01 已提交
3560 3561 3562 3563 3564 3565 3566 3567
        super(LstmLayer, self).__init__(name, 'lstmemory', 0, inputs, **xargs)
        config_assert(len(self.inputs) == 1, 'LstmLayer must have 1 input')
        input_layer = self.get_input_layer(0)
        #check input_layer.size is divided by 4
        config_assert(input_layer.size % 4 == 0, "size % 4 should be 0!")
        size = input_layer.size / 4
        self.set_layer_size(size)
        self.config.reversed = reversed
Q
qijun 已提交
3568
        self.config.active_gate_type = active_gate_type
Z
zhangjinchao01 已提交
3569 3570 3571 3572 3573
        self.config.active_state_type = active_state_type
        self.create_input_parameter(0, size * size * 4, [size, size, 4])
        #bias includes 3 kinds of peephole, 4 + 3 = 7
        self.create_bias_parameter(bias, size * 7)

Q
qijun 已提交
3574

Z
zhangjinchao01 已提交
3575 3576
@config_layer('lstm_step')
class LstmStepLayer(LayerBase):
Q
qijun 已提交
3577 3578 3579 3580 3581 3582 3583 3584 3585 3586
    def __init__(self,
                 name,
                 size,
                 inputs,
                 active_gate_type="sigmoid",
                 active_state_type="sigmoid",
                 bias=True,
                 **xargs):
        super(LstmStepLayer, self).__init__(name, 'lstm_step', size, inputs,
                                            **xargs)
Z
zhangjinchao01 已提交
3587 3588 3589
        config_assert(len(inputs) == 2, 'LstmStepLayer must have 2 inputs')
        input_layer0 = self.get_input_layer(0)
        input_layer1 = self.get_input_layer(1)
Q
qijun 已提交
3590 3591 3592 3593 3594
        config_assert(input_layer0.size == 4 * size,
                      'input_layer0.size != 4 * layer.size')
        config_assert(input_layer1.size == size,
                      'input_layer1.size != layer.size')
        self.config.active_gate_type = active_gate_type
Z
zhangjinchao01 已提交
3595 3596 3597
        self.config.active_state_type = active_state_type
        self.create_bias_parameter(bias, size * 3)

Q
qijun 已提交
3598

Z
zhangjinchao01 已提交
3599 3600 3601
# get the specific output from the input layer.
@config_layer('get_output')
class GetOutputLayer(LayerBase):
Q
qijun 已提交
3602 3603 3604 3605
    def __init__(self, name, size, inputs):
        super(GetOutputLayer, self).__init__(name, 'get_output', size, inputs)
        config_assert(
            len(self.inputs) == 1, 'GetOutputLayer must have 1 inputs')
Z
zhangjinchao01 已提交
3606 3607 3608 3609
        inputs = self.inputs[0]
        config_assert(inputs.input_layer_argument,
                      'input_layer_argument cannot be empty')

Q
qijun 已提交
3610

Z
zhangjinchao01 已提交
3611 3612
@config_layer('mdlstmemory')
class MDLstmLayer(LayerBase):
Q
qijun 已提交
3613 3614 3615 3616 3617 3618 3619 3620
    def __init__(self,
                 name,
                 inputs,
                 directions=True,
                 active_gate_type="sigmoid",
                 active_state_type="sigmoid",
                 bias=True,
                 **xargs):
Y
Yu Yang 已提交
3621 3622
        super(MDLstmLayer, self).__init__(name, 'mdlstmemory', 0, inputs,
                                          **xargs)
Z
zhangjinchao01 已提交
3623 3624 3625 3626
        config_assert(len(self.inputs) == 1, 'MDLstmLayer must have 1 input')
        input_layer = self.get_input_layer(0)
        dim_num = len(directions)
        #check input_layer.size is divided by (3+dim_num)
Y
Yu Yang 已提交
3627 3628
        config_assert(input_layer.size % (3 + dim_num) == 0,
                      "size % (dim_num) should be 0!")
Q
qijun 已提交
3629
        size = input_layer.size / (3 + dim_num)
Z
zhangjinchao01 已提交
3630
        self.set_layer_size(size)
Q
qijun 已提交
3631
        self.config.active_gate_type = active_gate_type
Z
zhangjinchao01 已提交
3632 3633 3634
        self.config.active_state_type = active_state_type
        for i in xrange(len(directions)):
            self.config.directions.append(int(directions[i]))
Y
Yu Yang 已提交
3635 3636
        self.create_input_parameter(0, size * size * (3 + dim_num),
                                    [size, size, 3 + dim_num])
Z
zhangjinchao01 已提交
3637
        #bias includes 3 kinds of peephole, 3+dim_num+2+dim_num
Q
qijun 已提交
3638 3639
        self.create_bias_parameter(bias, size * (5 + 2 * dim_num))

Z
zhangjinchao01 已提交
3640 3641 3642

@config_layer('gated_recurrent')
class GatedRecurrentLayer(LayerBase):
Q
qijun 已提交
3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653
    def __init__(self,
                 name,
                 inputs,
                 reversed=False,
                 active_gate_type="sigmoid",
                 bias=True,
                 **xargs):
        super(GatedRecurrentLayer, self).__init__(name, 'gated_recurrent', 0,
                                                  inputs, **xargs)
        config_assert(
            len(self.inputs) == 1, 'GatedRecurrentLayer must have 1 input')
Z
zhangjinchao01 已提交
3654 3655 3656 3657 3658 3659
        input_layer = self.get_input_layer(0)
        #check input_layer.size is divided by 3
        config_assert(input_layer.size % 3 == 0, "size % 3 should be 0!")
        size = input_layer.size / 3
        self.set_layer_size(size)
        self.config.reversed = reversed
Q
qijun 已提交
3660
        self.config.active_gate_type = active_gate_type
Z
zhangjinchao01 已提交
3661 3662 3663
        self.create_input_parameter(0, size * size * 3, [size, size * 3])
        self.create_bias_parameter(bias, size * 3)

Q
qijun 已提交
3664

Z
zhangjinchao01 已提交
3665 3666
@config_layer('gru_step')
class GruStepLayer(LayerBase):
Q
qijun 已提交
3667 3668 3669 3670 3671 3672 3673
    def __init__(self,
                 name,
                 size,
                 inputs,
                 active_gate_type="sigmoid",
                 bias=True,
                 **xargs):
Y
Yu Yang 已提交
3674 3675
        super(GruStepLayer, self).__init__(name, 'gru_step', size, inputs,
                                           **xargs)
Z
zhangjinchao01 已提交
3676 3677 3678
        config_assert(len(self.inputs) == 2, 'GruStepLayer must have 2 input')
        input_layer0 = self.get_input_layer(0)
        input_layer1 = self.get_input_layer(1)
Q
qijun 已提交
3679 3680 3681 3682 3683
        config_assert(input_layer0.size == 3 * size,
                      'input_layer0.size != 3 * layer.size')
        config_assert(input_layer1.size == size,
                      'input_layer1.size != layer.size')
        self.config.active_gate_type = active_gate_type
H
Haonan 已提交
3684
        self.create_input_parameter(0, size * size * 3, [size, size * 3])
Z
zhangjinchao01 已提交
3685 3686
        self.create_bias_parameter(bias, size * 3)

Q
qijun 已提交
3687

Z
zhangjinchao01 已提交
3688 3689 3690 3691 3692 3693 3694
'''
 A layer for calculating the cost of sequential conditional random field model.
 Example: CRFLayer(name="crf_cost", size=label_num,
                   inputs=["output", "label", "weight"])
          where "weight" is optional, one weight for each sequence
 @param coeff: weight of the layer
'''
Q
qijun 已提交
3695 3696


Z
zhangjinchao01 已提交
3697 3698
@config_layer('crf')
class CRFLayer(LayerBase):
Q
qijun 已提交
3699
    def __init__(self, name, size, inputs, coeff=1.0, device=None):
Z
zhangjinchao01 已提交
3700
        super(CRFLayer, self).__init__(name, 'crf', size, inputs, device=device)
Q
qijun 已提交
3701 3702
        config_assert(2 <= len(self.inputs) <= 3,
                      'CRFLayer must have 2 or 3 inputs')
3703
        self.create_input_parameter(0, size * (size + 2), [size + 2, size])
Z
zhangjinchao01 已提交
3704 3705
        self.config.coeff = coeff

Q
qijun 已提交
3706

Z
zhangjinchao01 已提交
3707 3708 3709 3710 3711 3712 3713 3714
'''
 A layer for calculating the decoding sequence of sequential conditional
 random field model.
 The decoding sequence is stored in output_.ids
 If a second input is provided, it is treated as the ground-truth label, and
 this layer will also calculate error, output_.value[i] is 1 for incorrect
 decoding or 0 for correct decoding
'''
Q
qijun 已提交
3715 3716


Z
zhangjinchao01 已提交
3717 3718
@config_layer('crf_decoding')
class CRFDecodingLayer(LayerBase):
Q
qijun 已提交
3719
    def __init__(self, name, size, inputs, device=None):
Z
zhangjinchao01 已提交
3720 3721 3722 3723 3724
        super(CRFDecodingLayer, self).__init__(
            name, 'crf_decoding', size, inputs, device=device)
        config_assert(
            len(self.inputs) <= 2,
            'CRFDecodingLayer cannot have more than 2 inputs')
3725
        self.create_input_parameter(0, size * (size + 2), [size + 2, size])
Z
zhangjinchao01 已提交
3726

Q
qijun 已提交
3727

Z
zhangjinchao01 已提交
3728 3729
@config_layer('ctc')
class CTCLayer(LayerBase):
Q
qijun 已提交
3730
    def __init__(self, name, size, inputs, norm_by_times=False, device=None):
Z
zhangjinchao01 已提交
3731 3732 3733 3734
        super(CTCLayer, self).__init__(name, 'ctc', size, inputs, device=device)
        self.config.norm_by_times = norm_by_times
        config_assert(len(self.inputs) == 2, 'CTCLayer must have 2 inputs')

Q
qijun 已提交
3735

3736 3737 3738 3739 3740 3741 3742 3743 3744 3745
@config_layer('kmax_seq_score')
class KmaxSeqScoreLayer(LayerBase):
    def __init__(self, name, inputs, beam_size, **xargs):
        super(KmaxSeqScoreLayer, self).__init__(
            name, 'kmax_seq_score', 0, inputs=inputs, **xargs)
        config_assert(
            len(self.inputs) == 1, 'KmaxSeqScoreLayer has only one input.')
        self.config.beam_size = beam_size


3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766
@config_layer('warp_ctc')
class WarpCTCLayer(LayerBase):
    def __init__(self,
                 name,
                 size,
                 inputs,
                 blank=0,
                 norm_by_times=False,
                 device=None):
        super(WarpCTCLayer, self).__init__(
            name, 'warp_ctc', size=size, inputs=inputs, device=device)
        self.config.blank = blank
        self.config.norm_by_times = norm_by_times
        config_assert(len(self.inputs) == 2, 'WarpCTCLayer must have 2 inputs')
        input_layer = self.get_input_layer(0)
        config_assert(
            (input_layer.active_type == '' or
             input_layer.active_type == 'linear'),
            "Expecting the active_type of input layer to be linear or null")


Z
zhangjinchao01 已提交
3767 3768
@config_layer('recurrent_layer_group')
class RecurrentLayerGroup(LayerBase):
Q
qijun 已提交
3769
    def __init__(self, name, device=None):
Z
zhangjinchao01 已提交
3770 3771 3772 3773
        super(RecurrentLayerGroup, self).__init__(
            name, 'recurrent_layer_group', 0, inputs=[], device=device)


3774 3775 3776 3777 3778
@config_layer('switch_order')
class SwitchOrderLayer(LayerBase):
    def __init__(self, name, inputs, reshape, **xargs):
        super(SwitchOrderLayer, self).__init__(
            name, 'switch_order', 0, inputs=inputs, **xargs)
W
wanghaoshuang 已提交
3779 3780
        self.config.reshape_conf.height_axis.extend(reshape['height'])
        self.config.reshape_conf.width_axis.extend(reshape['width'])
3781 3782


Z
zhangjinchao01 已提交
3783 3784
# Deprecated, use a new layer specific class instead
@config_func
Q
qijun 已提交
3785
def Layer(name, type, **xargs):
Z
zhangjinchao01 已提交
3786 3787 3788 3789
    layers = {}
    layers.update(g_cost_map)
    layers.update(g_layer_type_map)
    layer_func = layers.get(type)
Q
qijun 已提交
3790
    config_assert(layer_func, "layer type '%s' not supported." % type)
X
xuwei06 已提交
3791
    return layer_func(name, **xargs)
Z
zhangjinchao01 已提交
3792

Q
qijun 已提交
3793

Z
zhangjinchao01 已提交
3794
@config_func
Q
qijun 已提交
3795
def ParameterHook(type, **kwargs):
3796
    if type == 'pruning':
Z
zhangjinchao01 已提交
3797 3798
        hook = ParameterUpdaterHookConfig()
        hook.type = type
X
xzl 已提交
3799
        sparsity_ratio = kwargs.get('sparsity_ratio', None)
X
xzl 已提交
3800 3801
        if sparsity_ratio is not None:
            hook.sparsity_ratio = sparsity_ratio
Z
zhangjinchao01 已提交
3802
        return hook
3803 3804 3805 3806
    elif type == 'dpruning':
        hook = ParameterUpdaterHookConfig()
        hook.type = type
        return hook
Z
zhangjinchao01 已提交
3807 3808 3809 3810 3811
    else:
        return None


@config_func
Q
qijun 已提交
3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832
def Parameter(name,
              size,
              device,
              dims,
              learning_rate=None,
              momentum=None,
              decay_rate=None,
              decay_rate_l1=None,
              initial_mean=None,
              initial_std=None,
              initial_strategy=None,
              initial_smart=None,
              num_batches_regularization=None,
              sparse_remote_update=None,
              sparse_update=None,
              gradient_clipping_threshold=None,
              sparse=None,
              format=None,
              need_compact=None,
              is_static=None,
              is_shared=None,
X
xuwei06 已提交
3833 3834
              update_hooks=None,
              initializer=None):
Z
zhangjinchao01 已提交
3835 3836 3837 3838 3839 3840 3841

    config_assert(name not in g_parameter_map,
                  'Duplicated parameter name: ' + name)

    para = g_config.model_config.parameters.add()
    para.name = name
    para.size = size
3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852
    if device is not None:
        para.device = int(device)
    para.dims.extend(dims)

    if learning_rate is not None:
        para.learning_rate = float(learning_rate)

    momentum = default(momentum, g_default_momentum)
    if momentum is not None:
        para.momentum = float(momentum)

Z
zhangjinchao01 已提交
3853 3854
    config_assert(not momentum or not decay_rate_l1,
                  "momentum and decay_rate_l1 cannot both be non-zero")
3855 3856 3857 3858 3859

    decay_rate = default(decay_rate, g_default_decay_rate)
    if decay_rate is not None:
        para.decay_rate = decay_rate

Z
zhangjinchao01 已提交
3860 3861 3862 3863
    if decay_rate_l1 is not None:
        para.decay_rate_l1 = decay_rate_l1
    para.initial_std = default(initial_std, g_default_initial_std)
    para.initial_mean = default(initial_mean, g_default_initial_mean)
3864

Q
qijun 已提交
3865 3866
    num_batches_regularization = default(num_batches_regularization,
                                         g_default_num_batches_regularization)
3867 3868 3869
    if num_batches_regularization is not None:
        para.num_batches_regularization = int(num_batches_regularization)

Z
zhangjinchao01 已提交
3870 3871 3872 3873 3874 3875
    if sparse_remote_update is not None:
        para.sparse_remote_update = sparse_remote_update
        if sparse_remote_update:
            g_config.opt_config.use_sparse_remote_updater = True
    if sparse_update is not None:
        para.sparse_update = sparse_update
Q
qijun 已提交
3876 3877
    gradient_clipping_threshold = default(gradient_clipping_threshold,
                                          g_default_gradient_clipping_threshold)
3878 3879
    if gradient_clipping_threshold is not None:
        para.gradient_clipping_threshold = gradient_clipping_threshold
Q
qijun 已提交
3880 3881
    para.initial_strategy = default(initial_strategy,
                                    g_default_initial_strategy)
Z
zhangjinchao01 已提交
3882 3883 3884 3885 3886 3887
    para.initial_smart = default(initial_smart, g_default_initial_smart)
    if para.initial_smart:
        para.initial_mean = 0.
        if len(para.dims) != 0:
            para.initial_std = 1. / math.sqrt(para.dims[0])
        else:
Q
qijun 已提交
3888 3889 3890
            print(
                "Use initial_smart, but dims not set. Initial_smart may not be used in this layer"
            )
Z
zhangjinchao01 已提交
3891 3892 3893 3894
            traceback.print_exc()
            para.initial_std = 1. / math.sqrt(para.size)
    if g_default_compact_func is not None:
        sparse, format, need_compact = g_default_compact_func(para.name)
3895 3896 3897 3898 3899 3900 3901

    if sparse is not None:
        para.is_sparse = sparse
    if format is not None:
        para.format = format
    if need_compact is not None:
        para.need_compact = need_compact
Z
zhangjinchao01 已提交
3902 3903 3904 3905
    if is_static is not None:
        para.is_static = is_static
    config_assert(not para.sparse_remote_update or not para.is_static,
                  "sparse_remote_update and is_static cannot both be true")
3906 3907
    if is_shared is not None:
        para.is_shared = is_shared
Z
zhangjinchao01 已提交
3908 3909 3910 3911 3912

    update_hooks = default(update_hooks, g_default_update_hooks)

    if update_hooks is not None:
        if hasattr(update_hooks, '__call__'):
X
xzl 已提交
3913
            update_hooks = update_hooks()
Z
zhangjinchao01 已提交
3914 3915 3916 3917 3918

        if isinstance(update_hooks, list):
            for hook in update_hooks:
                para.update_hooks.extend([hook])
        else:
X
xzl 已提交
3919
            para.update_hooks.extend([update_hooks])
Z
zhangjinchao01 已提交
3920 3921

    g_parameter_map[name] = para
X
xuwei06 已提交
3922 3923 3924 3925 3926
    if initializer is not None:
        config_assert(
            callable(initializer),
            "parameter initializer should be a callable object")
        g_parameter_initializer_map[name] = initializer
Z
zhangjinchao01 已提交
3927 3928 3929 3930 3931 3932 3933


@config_func
def default_initial_std(val):
    global g_default_initial_std
    g_default_initial_std = val

Q
qijun 已提交
3934

Z
zhangjinchao01 已提交
3935 3936 3937 3938 3939
@config_func
def default_initial_mean(val):
    global g_default_initial_mean
    g_default_initial_mean = val

Q
qijun 已提交
3940

Z
zhangjinchao01 已提交
3941 3942 3943 3944 3945
@config_func
def default_initial_strategy(val):
    global g_default_initial_strategy
    g_default_initial_strategy = val

Q
qijun 已提交
3946

Z
zhangjinchao01 已提交
3947 3948 3949 3950 3951
@config_func
def default_initial_smart(val):
    global g_default_initial_smart
    g_default_initial_smart = val

Q
qijun 已提交
3952

Z
zhangjinchao01 已提交
3953 3954 3955 3956 3957
@config_func
def default_momentum(val):
    global g_default_momentum
    g_default_momentum = val

Q
qijun 已提交
3958

Z
zhangjinchao01 已提交
3959 3960 3961 3962 3963
@config_func
def default_decay_rate(val):
    global g_default_decay_rate
    g_default_decay_rate = val

Q
qijun 已提交
3964

Z
zhangjinchao01 已提交
3965 3966 3967 3968 3969
@config_func
def default_num_batches_regularization(val):
    global g_default_num_batches_regularization
    g_default_num_batches_regularization = val

Q
qijun 已提交
3970

Z
zhangjinchao01 已提交
3971 3972 3973 3974 3975
@config_func
def default_gradient_clipping_threshold(val):
    global g_default_gradient_clipping_threshold
    g_default_gradient_clipping_threshold = val

Q
qijun 已提交
3976

Z
zhangjinchao01 已提交
3977 3978 3979 3980 3981
@config_func
def default_device(val):
    global g_default_device
    g_default_device = val

Q
qijun 已提交
3982

Z
zhangjinchao01 已提交
3983 3984 3985 3986 3987
@config_func
def default_update_hooks(val):
    global g_default_update_hooks
    g_default_update_hooks = val

Q
qijun 已提交
3988

Z
zhangjinchao01 已提交
3989 3990 3991 3992 3993
@config_func
def default_compact_func(val):
    global g_default_compact_func
    g_default_compact_func = val

Q
qijun 已提交
3994

Z
zhangjinchao01 已提交
3995 3996 3997 3998 3999
def make_importer(config_dir, config_args):
    def Import(config_file, local_args={}):
        if not config_file.startswith('/'):
            config_file = config_dir + '/' + config_file
            g_config.config_files.append(config_file)
Q
qijun 已提交
4000 4001 4002
        execfile(config_file,
                 make_config_environment(config_file, config_args), local_args)

Z
zhangjinchao01 已提交
4003 4004
    return Import

Q
qijun 已提交
4005

X
xuwei06 已提交
4006
DEFAULT_SETTING = dict(
Z
zhangjinchao01 已提交
4007 4008 4009 4010 4011
    batch_size=None,
    mini_batch_size=None,
    algorithm='async_sgd',
    async_lagged_grad_discard_ratio=1.5,
    learning_method='momentum',
4012
    gradient_clipping_threshold=None,
Z
zhangjinchao01 已提交
4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034
    num_batches_per_send_parameter=None,
    num_batches_per_get_parameter=None,
    center_parameter_update_method=None,
    learning_rate=1.,
    learning_rate_decay_a=0.,
    learning_rate_decay_b=0.,
    learning_rate_schedule='poly',
    learning_rate_args='',
    l1weight=0.1,
    l2weight=0.,
    l2weight_zero_iter=0,
    c1=0.0001,
    backoff=0.5,
    owlqn_steps=10,
    max_backoff=5,
    average_window=0,
    do_average_in_cpu=False,
    max_average_window=None,
    ada_epsilon=1e-6,
    ada_rou=0.95,
    delta_add_rate=1.0,
    shrink_parameter_value=0,
Q
qijun 已提交
4035 4036 4037
    adam_beta1=0.9,
    adam_beta2=0.999,
    adam_epsilon=1e-8, )
Z
zhangjinchao01 已提交
4038

X
xuwei06 已提交
4039
settings = copy.deepcopy(DEFAULT_SETTING)
X
xuwei06 已提交
4040

Q
qijun 已提交
4041
settings_deprecated = dict(usage_ratio=1., )
Z
zhangjinchao01 已提交
4042 4043 4044 4045

trainer_settings = dict(
    save_dir="./output/model",
    init_model_path=None,
Q
qijun 已提交
4046 4047
    start_pass=0, )

Z
zhangjinchao01 已提交
4048 4049 4050 4051 4052

@config_func
def Settings(**args):
    for k, v in args.iteritems():
        if k == "usage_ratio":
Q
qijun 已提交
4053 4054
            logger.warning(
                "Deprecated: define usage_ratio in DataConfig instead")
Z
zhangjinchao01 已提交
4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065
            if g_config.HasField("data_config"):
                g_config.data_config.__setattr__(k, v)
            settings_deprecated[k] = v
            continue
        elif k in settings:
            settings[k] = v
        elif k in trainer_settings:
            trainer_settings[k] = v
        else:
            logger.fatal('Unkown setting: %s' % k)

Q
qijun 已提交
4066

Z
zhangjinchao01 已提交
4067 4068 4069 4070
@config_func
def cluster_config(**args):
    pass

Q
qijun 已提交
4071

Z
zhangjinchao01 已提交
4072 4073 4074 4075 4076 4077 4078 4079 4080
@config_func
def EnableSubmodelSuffix(flag=True):
    """
    If enabled, the layer and evaluator names in submodel will be automatically
    appended with @submodel_name
    """
    global g_add_submodel_suffix
    g_add_submodel_suffix = flag

Q
qijun 已提交
4081

Z
zhangjinchao01 已提交
4082 4083 4084 4085
def make_config_environment(config_file, config_args):
    def make_setter(k):
        def setter(v):
            logger.fatal("Obsolete: use Settings(%s=%s, ...) instead" % (k, v))
Q
qijun 已提交
4086

Z
zhangjinchao01 已提交
4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101
        return setter

    funcs = {}
    funcs.update(g_config_funcs)

    for k in settings.iterkeys():
        funcs[k] = make_setter(k)
    for k in settings_deprecated.iterkeys():
        funcs[k] = make_setter(k)
    config_dir = os.path.dirname(config_file)
    if not config_dir:
        config_dir = '.'

    funcs.update(
        Import=make_importer(config_dir, config_args),
Q
qijun 已提交
4102
        get_config_arg=make_get_config_arg(config_args), )
Z
zhangjinchao01 已提交
4103 4104 4105 4106 4107

    funcs.update(g_extended_config_funcs)

    return funcs

Q
qijun 已提交
4108

Z
zhangjinchao01 已提交
4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124
def make_get_config_arg(config_args):
    def get_config_arg(name, type, default=None):
        if type == bool:
            s = config_args.get(name)
            if not s:
                return default
            if s == 'True' or s == '1' or s == 'true':
                return True
            if s == 'False' or s == '0' or s == 'false':
                return False
            raise ValueError('Value of config_arg %s is not boolean' % name)
        else:
            return type(config_args.get(name, default))

    return get_config_arg

Q
qijun 已提交
4125

Z
zhangjinchao01 已提交
4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137
def importlib(name):
    __import__(name)
    return sys.modules[name]


def find_caller():
    stack = traceback.extract_stack()
    for s in stack[-4::-1]:
        if not s[0].endswith('config_parser.py'):
            return s[0], s[1], s[2]
    return "(unknown file)", 0, "(unknown function)"

Q
qijun 已提交
4138

Z
zhangjinchao01 已提交
4139 4140 4141 4142
def my_fatal(s):
    logger.critical(s)
    raise Exception()

Y
Yu Yang 已提交
4143

4144
_parse_config_hooks = set()
Y
Yu Yang 已提交
4145 4146


4147 4148 4149 4150 4151 4152 4153
def register_parse_config_hook(f):
    """
    Register a hook function for parse_config. parse_config will invoke the hook
    at the beginning of parse. This make it possible to reset global state for
    for constructing the model.
    """
    _parse_config_hooks.add(f)
Q
qijun 已提交
4154

Y
Yu Yang 已提交
4155

4156
def update_g_config():
Z
zhangjinchao01 已提交
4157
    '''
4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180
    Update g_config after execute config_file or config_functions.
    '''
    for k, v in settings.iteritems():
        if v is None:
            continue
        g_config.opt_config.__setattr__(k, v)

    for k, v in trainer_settings.iteritems():
        if v is None:
            continue
        g_config.__setattr__(k, v)

    for name in g_config.model_config.input_layer_names:
        assert name in g_layer_map, \
            'input name "%s" does not correspond to a layer name' % name
        assert (g_layer_map[name].type == "data" or g_layer_map[name].type == "data_trim"), \
            'The type of input layer "%s" is not "data"' % name
    for name in g_config.model_config.output_layer_names:
        assert name in g_layer_map, \
            'input name "%s" does not correspond to a layer name' % name
    return g_config


4181
def begin_parse():
Z
zhangjinchao01 已提交
4182
    init_config_environment()
4183 4184
    for hook in _parse_config_hooks:
        hook()
Z
zhangjinchao01 已提交
4185 4186 4187 4188 4189

    logger.findCaller = find_caller
    logger.fatal = my_fatal

    g_config.model_config.type = "nn"
X
xuwei06 已提交
4190 4191 4192 4193 4194 4195 4196 4197 4198

    global g_current_submodel, g_root_submodel
    g_root_submodel = g_config.model_config.sub_models.add()
    g_root_submodel.name = 'root'
    g_root_submodel.is_recurrent_layer_group = False
    g_current_submodel = g_root_submodel


def parse_config(trainer_config, config_arg_str):
4199 4200 4201 4202
    '''
    @param config_arg_str: a string of the form var1=val1,var2=val2. It will be
    passed to config script as a dictionary CONFIG_ARGS
    '''
X
xuwei06 已提交
4203

4204
    begin_parse()
X
xuwei06 已提交
4205 4206
    config_args = {}

Z
zhangjinchao01 已提交
4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218
    if config_arg_str:
        config_args = dict([f.split('=') for f in config_arg_str.split(',')])

    global g_command_config_args
    g_command_config_args.update(config_args)

    extension_module_name = config_args.get('extension_module_name')
    if extension_module_name:
        global g_extended_config_funcs
        extension_module = importlib(extension_module_name)
        g_extended_config_funcs = extension_module.get_config_funcs(g_config)

4219 4220
    if hasattr(trainer_config, '__call__'):
        trainer_config.func_globals.update(
L
Luo Tao 已提交
4221
            make_config_environment("", config_args))
4222
        trainer_config()
H
hanchao 已提交
4223
    else:
4224 4225
        execfile(trainer_config,
                 make_config_environment(trainer_config, config_args))
Z
zhangjinchao01 已提交
4226

4227
    return update_g_config()
Z
zhangjinchao01 已提交
4228 4229


4230
def parse_config_and_serialize(trainer_config, config_arg_str):
Z
zhangjinchao01 已提交
4231
    try:
4232
        config = parse_config(trainer_config, config_arg_str)
Z
zhangjinchao01 已提交
4233 4234 4235 4236 4237 4238
        #logger.info(config)
        return config.SerializeToString()
    except:
        traceback.print_exc()
        raise

Q
qijun 已提交
4239

Z
zhangjinchao01 已提交
4240 4241 4242 4243 4244 4245 4246 4247
if __name__ == '__main__':
    try:
        config = parse_config(sys.argv[1], '')
        config.SerializeToString()
        __real_print__(str(config))
    except:
        traceback.print_exc()
        raise