config_parser.py 135.4 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
        ScatterAgentLayer(name=name, size=layer.size)
342

Z
zhangjinchao01 已提交
343 344 345 346
        pair = g_current_submodel.in_links.add()
        pair.layer_name = layer_name
        pair.link_name = MakeLayerNameInSubmodel(name)

Q
qijun 已提交
347

Z
zhangjinchao01 已提交
348 349 350 351 352 353 354 355 356 357 358 359 360
@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
361
    generator.eos_layer_name = MakeLayerNameInSubmodel(generator.eos_layer_name)
Z
zhangjinchao01 已提交
362 363 364 365 366 367 368 369
    g_current_submodel.generator.CopyFrom(generator)


@config_func
def RecurrentLayerGroupBegin(name,
                             in_links,
                             out_links,
                             generator=None,
370
                             target_inlinkname="",
Z
zhangjinchao01 已提交
371
                             seq_reversed=False):
372
    RecurrentLayerGroupWithoutOutLinksBegin(name, in_links, seq_reversed)
Z
zhangjinchao01 已提交
373 374 375 376 377
    for link in out_links:
        RecurrentLayerGroupSetOutLink(link)

    if generator is not None:
        RecurrentLayerGroupSetGenerator(generator)
Q
qijun 已提交
378 379 380 381 382
        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
383 384 385 386 387 388 389


@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
390
    for pair in g_current_submodel.memories:  #check exist
Z
zhangjinchao01 已提交
391
        layer = g_layer_map[pair.layer_name]
Y
Yu Yang 已提交
392 393
        config_assert(layer is not None,
                      "memory declare wrong name:%s" % pair.layer_name)
Z
zhangjinchao01 已提交
394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409
        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
410

Z
zhangjinchao01 已提交
411 412 413 414 415 416
# 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
417

Z
zhangjinchao01 已提交
418 419
@config_class
class Bias(Cfg):
X
xuwei06 已提交
420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435
    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
436 437
        self.add_keys(locals())

Q
qijun 已提交
438

Z
zhangjinchao01 已提交
439 440 441 442 443 444 445
# 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
446
            initializer=None,
Z
zhangjinchao01 已提交
447 448 449 450 451 452 453 454 455 456 457 458 459
            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
460
            bilinear_interp=None,
Z
zhangjinchao01 已提交
461 462 463 464
            norm=None,
            pool=None,
            image=None,
            block_expand=None,
465
            maxout=None,
Q
qijun 已提交
466
            spp=None,
D
dangqingqing 已提交
467
            pad=None,
Z
zhangjinchao01 已提交
468 469 470 471 472
            format=None,
            nnz=None,
            is_static=None,
            is_shared=None,
            update_hooks=None,
473
            input_layer_argument=None,
D
dangqingqing 已提交
474 475 476 477 478
            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
479
        self.add_keys(locals())
D
dangqingqing 已提交
480 481 482
        self.input_layer_name = MakeLayerNameInSubmodel(
            input_layer_name
        ) if make_layer_name_in_submodel else input_layer_name
Z
zhangjinchao01 已提交
483

Q
qijun 已提交
484

Z
zhangjinchao01 已提交
485 486 487
# Define a projection for iexed layer
@config_class
class Projection(Input):
Q
qijun 已提交
488 489
    type = None  # subclass should set it correctly

Z
zhangjinchao01 已提交
490 491 492
    def __init__(
            self,
            input_layer_name,
Q
qijun 已提交
493
            size=0,  # projection output size
Z
zhangjinchao01 已提交
494 495 496 497 498 499 500 501 502
            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
503
            initializer=None,
Z
zhangjinchao01 已提交
504 505 506 507 508 509 510 511 512 513
            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
514
            input_layer_argument=None, ):
Z
zhangjinchao01 已提交
515 516 517 518 519 520 521 522 523 524 525 526 527
        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
528

Z
zhangjinchao01 已提交
529 530
    def calc_parameter_size(self, input_size, output_size):
        raise NotimplementedError
Q
qijun 已提交
531

Z
zhangjinchao01 已提交
532 533 534 535 536 537 538 539 540 541
    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
542

Z
zhangjinchao01 已提交
543 544
    def calc_parameter_size(self, input_size, output_size):
        return 0
Q
qijun 已提交
545

Z
zhangjinchao01 已提交
546 547 548
    def calc_parameter_dims(self, input_size, output_size):
        return []

Q
qijun 已提交
549

Z
zhangjinchao01 已提交
550 551 552 553 554 555
# 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
556 557 558
    def __init__(self, input_layer_name, offset, **xargs):
        super(IdentityOffsetProjection, self).__init__(input_layer_name,
                                                       **xargs)
Z
zhangjinchao01 已提交
559 560 561 562
        self.proj_conf.offset = offset

    def calc_parameter_size(self, input_size, output_size):
        return 0
Q
qijun 已提交
563

Z
zhangjinchao01 已提交
564 565 566
    def calc_parameter_dims(self, input_size, output_size):
        return []

Q
qijun 已提交
567

568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596
@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
597 598 599 600 601 602 603
# 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
604

Z
zhangjinchao01 已提交
605 606
    def calc_parameter_size(self, input_size, output_size):
        return output_size
Q
qijun 已提交
607

Z
zhangjinchao01 已提交
608 609 610
    def calc_parameter_dims(self, input_size, output_size):
        return [1, output_size]

L
Luo Tao 已提交
611

X
xuwei06 已提交
612 613 614 615 616 617 618 619 620 621 622 623 624 625
# 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
626

Z
zhangjinchao01 已提交
627 628 629 630 631 632
@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
633

Z
zhangjinchao01 已提交
634 635 636
    def calc_parameter_dims(self, input_size, output_size):
        return [input_size, output_size]

Q
qijun 已提交
637

Z
zhangjinchao01 已提交
638 639 640 641 642 643
@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
644

Z
zhangjinchao01 已提交
645 646 647
    def calc_parameter_dims(self, input_size, output_size):
        return [input_size, output_size]

Q
qijun 已提交
648

Z
zhangjinchao01 已提交
649 650 651 652 653 654
@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
655

Z
zhangjinchao01 已提交
656 657 658
    def calc_parameter_dims(self, input_size, output_size):
        return [output_size, input_size]

Q
qijun 已提交
659

Z
zhangjinchao01 已提交
660 661 662 663
@config_class
class ContextProjection(Projection):
    type = 'context'

Q
qijun 已提交
664 665
    def __init__(self, input_layer_name, context_start, context_length,
                 trainable_padding, **xargs):
Z
zhangjinchao01 已提交
666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688
        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


689
@config_class
690
class ConvBaseProjection(Projection):
Q
qijun 已提交
691 692 693 694 695
    def __init__(self,
                 input_layer_name,
                 num_filters=None,
                 conv_conf=None,
                 **xargs):
696
        super(ConvBaseProjection, self).__init__(input_layer_name, **xargs)
697 698 699 700 701 702 703 704 705 706 707 708

        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
709 710
        gr = self.proj_conf.conv_conf.groups
        return co * ci * fh * fw / gr
711 712 713 714 715 716 717

    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
718

719 720 721 722 723 724 725 726 727
@config_class
class ConvProjection(ConvBaseProjection):
    type = 'conv'

    def __init__(self,
                 input_layer_name,
                 num_filters=None,
                 conv_conf=None,
                 **xargs):
728 729
        super(ConvProjection, self).__init__(input_layer_name, num_filters,
                                             conv_conf, **xargs)
730

731
        parse_conv(conv_conf, self.input_layer_name, self.proj_conf.conv_conf,
732 733 734 735 736 737 738 739 740 741 742 743 744 745 746
                   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):
747 748
        super(ConvTransProjection, self).__init__(input_layer_name, num_filters,
                                                  conv_conf, **xargs)
749 750 751

        parse_conv(
            conv_conf,
752
            self.input_layer_name,
753 754 755 756 757 758 759 760
            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
761 762 763
# Define a operator for mixed layer
@config_class
class Operator(Cfg):
Q
qijun 已提交
764 765
    type = None  # subclass should set it correctly

Z
zhangjinchao01 已提交
766 767
    def __init__(
            self,
Q
qijun 已提交
768
            input_layer_names, ):
Z
zhangjinchao01 已提交
769 770 771 772 773 774 775 776 777 778
        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
779

Z
zhangjinchao01 已提交
780 781 782
@config_class
class DotMulOperator(Operator):
    type = 'dot_mul'
Q
qijun 已提交
783 784 785

    def __init__(self, input_layer_names, scale=None, **xargs):
        super(DotMulOperator, self).__init__(input_layer_names, **xargs)
Z
zhangjinchao01 已提交
786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803
        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
804 805 806 807 808 809 810

    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
811 812 813
        if num_filters is not None:
            self.operator_conf.num_filters = num_filters

814 815
        parse_conv(conv_conf,
                   MakeLayerNameInSubmodel(input_layer_names[0]),
Q
qijun 已提交
816
                   self.operator_conf.conv_conf, num_filters)
L
Luo Tao 已提交
817 818 819
        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
820 821 822

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

823 824
    def calc_output_size(self, input_sizes):
        return self.operator_conf.output_size
Z
zhangjinchao01 已提交
825 826


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
@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
857 858 859
# please refer to the comments in proto/ModelConfig.proto
@config_class
class Conv(Cfg):
Q
qijun 已提交
860 861 862 863 864 865 866 867 868 869 870 871 872
    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,
                 stride_y=None):
Z
zhangjinchao01 已提交
873 874
        self.add_keys(locals())
        if filter_size_y is None:
Q
qijun 已提交
875
            self.filter_size_y = filter_size
Z
zhangjinchao01 已提交
876
        if padding_y is None:
Q
qijun 已提交
877
            self.padding_y = padding
Z
zhangjinchao01 已提交
878
        if stride_y is None:
Q
qijun 已提交
879
            self.stride_y = stride
Z
zhangjinchao01 已提交
880
        if output_x is not None:
Q
qijun 已提交
881 882
            config_assert(output_x <= 0)

Z
zhangjinchao01 已提交
883

L
liaogang 已提交
884 885
@config_class
class BilinearInterp(Cfg):
L
Luo Tao 已提交
886
    def __init__(self, out_size_x=None, out_size_y=None, channels=None):
L
liaogang 已提交
887 888
        self.add_keys(locals())

Q
qijun 已提交
889

Z
zhangjinchao01 已提交
890 891
@config_class
class Pool(Cfg):
D
dangqingqing 已提交
892 893 894 895 896 897 898 899 900 901 902
    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
903
        self.add_keys(locals())
Q
qijun 已提交
904 905


Q
qijun 已提交
906
@config_class
Q
qijun 已提交
907
class SpatialPyramidPool(Cfg):
L
Luo Tao 已提交
908
    def __init__(self, pool_type, pyramid_height, channels):
Q
qijun 已提交
909
        self.add_keys(locals())
Z
zhangjinchao01 已提交
910

Q
qijun 已提交
911

D
dangqingqing 已提交
912 913 914 915 916 917
@config_class
class Pad(Cfg):
    def __init__(self, channels, pad_c, pad_h, pad_w):
        self.add_keys(locals())


Z
zhangjinchao01 已提交
918 919
@config_class
class Norm(Cfg):
Q
qijun 已提交
920 921 922 923 924 925 926 927 928
    def __init__(self,
                 norm_type,
                 channels,
                 size,
                 scale,
                 pow,
                 output_x=None,
                 img_size=None,
                 blocked=None):
Z
zhangjinchao01 已提交
929 930
        self.add_keys(locals())

Q
qijun 已提交
931

Z
zhangjinchao01 已提交
932 933
@config_class
class Image(Cfg):
Q
qijun 已提交
934
    def __init__(self, channels, img_size=None):
Z
zhangjinchao01 已提交
935 936
        self.add_keys(locals())

Q
qijun 已提交
937

Z
zhangjinchao01 已提交
938 939
@config_class
class BlockExpand(Cfg):
Q
qijun 已提交
940 941 942 943 944 945 946 947 948 949 950 951
    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
952 953
        self.add_keys(locals())

Q
qijun 已提交
954

955 956
@config_class
class MaxOut(Cfg):
Q
qijun 已提交
957
    def __init__(self, channels, groups, img_size_x=0, img_size_y=0):
958 959
        self.add_keys(locals())

Q
qijun 已提交
960

961
def create_data_config_proto(async_load_data=False,
962
                             constant_slots=None,
王益 已提交
963 964 965
                             data_ratio=1,
                             is_main_data=True,
                             usage_ratio=None):
Z
zhangjinchao01 已提交
966 967 968 969 970 971 972 973
    # 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
974 975
    data_config.data_ratio = data_ratio
    data_config.is_main_data = is_main_data
Z
zhangjinchao01 已提交
976

Q
qijun 已提交
977
    usage_ratio = default(usage_ratio, settings_deprecated["usage_ratio"])
Z
zhangjinchao01 已提交
978 979 980 981 982 983
    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
984

Z
zhangjinchao01 已提交
985
@config_func
Q
qijun 已提交
986 987 988 989 990
def SimpleData(files=None,
               feat_dim=None,
               context_len=None,
               buffer_capacity=None,
               **xargs):
991
    data_config = create_data_config_proto(**xargs)
Z
zhangjinchao01 已提交
992 993 994 995 996 997 998 999 1000
    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
1001

Z
zhangjinchao01 已提交
1002
@config_func
Q
qijun 已提交
1003 1004 1005 1006 1007 1008 1009 1010 1011 1012
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):
1013
    data_config = create_data_config_proto(**xargs)
Z
zhangjinchao01 已提交
1014 1015
    data_config.type = 'py'
    if load_data_module in g_py_module_name_list:
Q
qijun 已提交
1016

Z
zhangjinchao01 已提交
1017 1018 1019
        def get_path(module):
            m = __import__(load_data_module)
            return os.path.split(os.path.realpath(m.__file__))[0]
Q
qijun 已提交
1020

Z
zhangjinchao01 已提交
1021 1022 1023
        # 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
1024 1025
        module_new_name = "%s_copy_%d" % (load_data_module,
                                          len(g_py_module_name_list))
Z
zhangjinchao01 已提交
1026
        g_py_module_name_list.append(module_new_name)
Q
qijun 已提交
1027 1028 1029 1030
        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
1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054
        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
1055

Z
zhangjinchao01 已提交
1056
@config_func
Q
qijun 已提交
1057 1058 1059 1060 1061 1062 1063
def ProtoData(files=None,
              type=None,
              file_group_queue_capacity=None,
              load_file_count=None,
              constant_slots=None,
              load_thread_num=None,
              **xargs):
1064
    data_config = create_data_config_proto(**xargs)
Z
zhangjinchao01 已提交
1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083
    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
1084

Z
zhangjinchao01 已提交
1085 1086
#real data for training is actually provided by "sub_data" data providers.
@config_func
Q
qijun 已提交
1087
def MultiData(sub_data=[]):
Z
zhangjinchao01 已提交
1088 1089 1090 1091 1092
    data_config = DataConfig()
    data_config.type = 'multi'
    data_config.sub_data_configs.extend(sub_data)
    return data_config

Q
qijun 已提交
1093

Z
zhangjinchao01 已提交
1094
@config_func
Q
qijun 已提交
1095 1096 1097 1098 1099 1100 1101
def Data(type,
         files=None,
         feat_dim=None,
         slot_dims=None,
         context_len=None,
         buffer_capacity=None,
         **xargs):
Z
zhangjinchao01 已提交
1102

1103
    data_config = create_data_config_proto(**xargs)
Z
zhangjinchao01 已提交
1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136
    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
1137

L
Luo Tao 已提交
1138 1139
#caffe_mode: compute the output size using floor instead of ceil,
#            which is consistent of caffe and CuDNN's convention.
1140 1141 1142 1143 1144 1145 1146
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
1147

1148
#calcualte image_size based on output_size for de-convolution (ConvTransLayer).
L
Luo Tao 已提交
1149
#It is the reverse function of cnn_output_size
1150
def cnn_image_size(output_size, filter_size, padding, stride, caffe_mode):
L
Luo Tao 已提交
1151 1152 1153
    img_size = (output_size - 1) * stride + filter_size - 2 * padding
    if not caffe_mode:
        img_size = img_size + 1
1154 1155
    return img_size

Q
qijun 已提交
1156

L
Luo Tao 已提交
1157
def get_img_size(input_layer_name, channels):
L
Luo Tao 已提交
1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175
    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


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


1176
def parse_pool(pool, input_layer_name, pool_conf, ceil_mode):
Z
zhangjinchao01 已提交
1177
    pool_conf.pool_type = pool.pool_type
Q
qijun 已提交
1178 1179 1180
    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
1181
                  "['max-projection', 'avg-projection', "
Q
qijun 已提交
1182
                  "'cudnn-max-pool', 'cudnn-avg-pool']" % pool.pool_type)
Z
zhangjinchao01 已提交
1183 1184 1185 1186 1187 1188

    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
1189
    pool_conf.stride_y = default(pool.stride_y, pool_conf.stride)
Z
zhangjinchao01 已提交
1190

L
Luo Tao 已提交
1191
    pool_conf.img_size, pool_conf.img_size_y = \
L
Luo Tao 已提交
1192
        get_img_size(input_layer_name, pool.channels)
Z
zhangjinchao01 已提交
1193

1194
    config_assert(not pool.start, "start is deprecated in pooling.")
Z
zhangjinchao01 已提交
1195

1196
    if pool.padding is not None:
Z
zhangjinchao01 已提交
1197
        pool_conf.padding = pool.padding
1198
    pool_conf.padding_y = default(pool.padding_y, pool_conf.padding)
D
dangqingqing 已提交
1199 1200
    pool_conf.output_x = cnn_output_size(pool_conf.img_size, pool_conf.size_x,
                                         pool_conf.padding, pool_conf.stride,
1201
                                         not ceil_mode)
D
dangqingqing 已提交
1202 1203
    pool_conf.output_y = cnn_output_size(pool_conf.img_size_y, pool_conf.size_y,
                                         pool_conf.padding_y,
1204
                                         pool_conf.stride_y, not ceil_mode)
Q
qijun 已提交
1205

Z
zhangjinchao01 已提交
1206

Q
qijun 已提交
1207
def parse_spp(spp, input_layer_name, spp_conf):
L
Luo Tao 已提交
1208
    parse_image(spp, input_layer_name, spp_conf.image_conf)
Q
qijun 已提交
1209 1210
    spp_conf.pool_type = spp.pool_type
    config_assert(spp.pool_type in ['max-projection', 'avg-projection'],
Q
qijun 已提交
1211 1212
                  "pool-type %s is not in "
                  "['max-projection', 'avg-projection']" % spp.pool_type)
Q
qijun 已提交
1213
    spp_conf.pyramid_height = spp.pyramid_height
Q
qijun 已提交
1214

Q
qijun 已提交
1215

Z
zhangjinchao01 已提交
1216 1217
def parse_image(image, input_layer_name, image_conf):
    image_conf.channels = image.channels
L
Luo Tao 已提交
1218
    image_conf.img_size, image_conf.img_size_y = \
L
Luo Tao 已提交
1219
        get_img_size(input_layer_name, image_conf.channels)
Q
qijun 已提交
1220

Z
zhangjinchao01 已提交
1221 1222 1223

def parse_norm(norm, input_layer_name, norm_conf):
    norm_conf.norm_type = norm.norm_type
1224 1225 1226 1227 1228
    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
1229 1230 1231 1232 1233 1234
    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
1235
    norm_conf.img_size, norm_conf.img_size_y = \
L
Luo Tao 已提交
1236
        get_img_size(input_layer_name, norm.channels)
Z
zhangjinchao01 已提交
1237
    norm_conf.output_x = norm_conf.img_size
L
Luo Tao 已提交
1238
    norm_conf.output_y = norm_conf.img_size_y
Z
zhangjinchao01 已提交
1239 1240 1241
    if norm.norm_type in ['cmrnorm-projection']:
        norm_conf.scale /= norm.size
    else:
Q
qijun 已提交
1242 1243
        norm_conf.scale /= norm.size**2

1244

L
Luo Tao 已提交
1245 1246
#caffe_mode: compute the output size using floor instead of ceil,
#            which is consistent of caffe and CuDNN's convention.
1247
def parse_conv(conv, input_layer_name, conv_conf, num_filters, trans=False):
Z
zhangjinchao01 已提交
1248 1249 1250 1251 1252 1253 1254 1255 1256
    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
1257

1258
    if not trans:
1259
        conv_conf.filter_channels = conv.channels / conv.groups
L
Luo Tao 已提交
1260
        conv_conf.img_size, conv_conf.img_size_y = \
L
Luo Tao 已提交
1261
            get_img_size(input_layer_name, conv.channels)
1262
        conv_conf.output_x = cnn_output_size(
Q
qijun 已提交
1263 1264
            conv_conf.img_size, conv_conf.filter_size, conv_conf.padding,
            conv_conf.stride, conv_conf.caffe_mode)
L
Luo Tao 已提交
1265 1266 1267
        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)
1268
    else:
1269
        conv_conf.filter_channels = num_filters / conv.groups
L
Luo Tao 已提交
1270
        conv_conf.output_x, conv_conf.output_y = \
L
Luo Tao 已提交
1271
            get_img_size(input_layer_name, conv.channels)
1272
        conv_conf.img_size = cnn_image_size(
Q
qijun 已提交
1273 1274
            conv_conf.output_x, conv_conf.filter_size, conv_conf.padding,
            conv_conf.stride, conv_conf.caffe_mode)
L
Luo Tao 已提交
1275
        conv_conf.img_size_y = cnn_image_size(
L
Luo Tao 已提交
1276 1277
            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
1278

1279

Z
zhangjinchao01 已提交
1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292
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:
1293
        block_expand_conf.output_x = cnn_output_size(
1294
            block_expand.img_size_x, block_expand.block_x,
1295
            block_expand.padding_x, block_expand.stride_x, False)
Z
zhangjinchao01 已提交
1296 1297

    if block_expand_conf.img_size_y == 0:
1298
        block_expand_conf.output_y = 0
Z
zhangjinchao01 已提交
1299
    else:
1300
        block_expand_conf.output_y = cnn_output_size(
1301
            block_expand.img_size_y, block_expand.block_y,
1302
            block_expand.padding_y, block_expand.stride_y, False)
Z
zhangjinchao01 已提交
1303

Q
qijun 已提交
1304

1305
def parse_maxout(maxout, input_layer_name, maxout_conf):
L
Luo Tao 已提交
1306
    parse_image(maxout, input_layer_name, maxout_conf.image_conf)
1307
    maxout_conf.groups = maxout.groups
1308

Q
qijun 已提交
1309

Z
zhangjinchao01 已提交
1310 1311
# Define an evaluator
@config_func
Y
yangyaming 已提交
1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328
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
1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342
    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)

1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353
    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
1354 1355
    if top_k is not None:
        evaluator.top_k = top_k
1356 1357
    if delimited is not None:
        evaluator.delimited = delimited
Z
zhangjinchao01 已提交
1358

1359 1360 1361
    if excluded_chunk_types:
        evaluator.excluded_chunk_types.extend(excluded_chunk_types)

Y
yangyaming 已提交
1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373
    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
1374

Z
zhangjinchao01 已提交
1375 1376 1377 1378 1379
class LayerBase(object):
    def __init__(
            self,
            name,
            type,
Q
qijun 已提交
1380
            size,  # size can be 0. In this case, subclass should set it.
Z
zhangjinchao01 已提交
1381 1382 1383 1384
            inputs,
            device=None,
            active_type="",
            drop_rate=0.,
C
caoying03 已提交
1385 1386
            coeff=None,
            error_clipping_threshold=None):
Z
zhangjinchao01 已提交
1387
        config_assert('@' not in name,
Q
qijun 已提交
1388
                      "layer name: %s contain special character @" % name)
Z
zhangjinchao01 已提交
1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403
        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()
1404
        assert isinstance(self.config, LayerConfig)
Z
zhangjinchao01 已提交
1405 1406 1407
        self.config.name = name
        self.config.type = type
        self.config.active_type = active_type
1408 1409
        if coeff is not None:
            self.config.coeff = float(coeff)
Z
zhangjinchao01 已提交
1410 1411 1412 1413 1414 1415 1416
        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
1417
        elif g_default_device is not None:
Z
zhangjinchao01 已提交
1418 1419
            self.config.device = g_default_device

C
caoying03 已提交
1420 1421 1422
        if error_clipping_threshold is not None:
            self.config.error_clipping_threshold = error_clipping_threshold

Z
zhangjinchao01 已提交
1423 1424 1425 1426 1427 1428 1429
        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
1430 1431
                    input_layer_name=input,
                    parameter_name=gen_parameter_name(name, input_index))
Z
zhangjinchao01 已提交
1432 1433 1434 1435 1436 1437 1438 1439
                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
1440
                self.operators.append(input)
Z
zhangjinchao01 已提交
1441 1442 1443 1444
                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
1445
                raise ValueError('Wrong type for inputs: %s' % type_of(input))
Z
zhangjinchao01 已提交
1446
            config_assert(input_layer_name in g_layer_map,
Q
qijun 已提交
1447 1448
                          "Unknown input layer '%s' for layer %s" %
                          (input_layer_name, name))
Z
zhangjinchao01 已提交
1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465
            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
1466
            bias,  # True/False or BiasCfg
Z
zhangjinchao01 已提交
1467
            size,
Q
qijun 已提交
1468 1469 1470
            dims=None,
            for_self=True,  # whether create bias for layer self
    ):
Z
zhangjinchao01 已提交
1471 1472 1473 1474 1475 1476

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

Q
qijun 已提交
1477 1478 1479
        config_assert(
            type_of(bias) == bool or type_of(bias) == Bias,
            'Incorrect type for bias: %s' % type_of(bias))
Z
zhangjinchao01 已提交
1480 1481 1482 1483 1484 1485 1486 1487 1488

        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:
1489 1490
                assert isinstance(self.config, LayerConfig)

Z
zhangjinchao01 已提交
1491 1492 1493
                Parameter(
                    bias.parameter_name,
                    size,
Q
qijun 已提交
1494 1495
                    self.config.device
                    if self.config.HasField('device') else None,
Z
zhangjinchao01 已提交
1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506
                    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
1507 1508
                    gradient_clipping_threshold=bias.
                    gradient_clipping_threshold,
Z
zhangjinchao01 已提交
1509
                    is_static=bias.is_static,
X
xuwei06 已提交
1510 1511
                    is_shared=bias.is_shared,
                    initializer=bias.initializer)
Z
zhangjinchao01 已提交
1512 1513 1514 1515 1516
            if for_self:
                self.config.bias_parameter_name = bias.parameter_name
            else:
                return bias.parameter_name

Q
qijun 已提交
1517 1518 1519 1520 1521 1522
    def create_input_parameter(self,
                               input_index,
                               size,
                               dims=None,
                               sparse=None,
                               format=None):
Z
zhangjinchao01 已提交
1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536
        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
1537 1538
            config_assert(size == para.size, (
                'Shared parameter "%s" does not ' + 'have same size: %s vs. %s')
Z
zhangjinchao01 已提交
1539 1540
                          % (input_config.parameter_name, para.size, size))

Q
qijun 已提交
1541 1542
            config_assert(dims == para.dims, (
                'Shared parameter "%s" does not ' + 'have same dims: %s vs. %s')
Z
zhangjinchao01 已提交
1543 1544 1545 1546 1547 1548
                          % (input_config.parameter_name, para.dims, dims))
            return

        Parameter(
            input_config.parameter_name,
            size,
1549
            self.config.device if self.config.HasField("device") else None,
Z
zhangjinchao01 已提交
1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561
            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
1562 1563
            gradient_clipping_threshold=input_config.
            gradient_clipping_threshold,
Z
zhangjinchao01 已提交
1564 1565 1566 1567
            sparse=sparse,
            format=format,
            is_static=input_config.is_static,
            is_shared=input_config.is_shared,
X
xuwei06 已提交
1568 1569
            update_hooks=input_config.update_hooks,
            initializer=input_config.initializer)
Z
zhangjinchao01 已提交
1570 1571 1572 1573 1574 1575 1576 1577 1578

    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
1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595
    def set_layer_height_width(self, height, width):
        self.config.height = height
        self.config.width = width

    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
1596

Z
zhangjinchao01 已提交
1597 1598
@config_layer('multi_class_cross_entropy_with_selfnorm')
class MultiClassCrossEntropySelfNormCostLayer(LayerBase):
Q
qijun 已提交
1599 1600 1601
    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
1602 1603
        self.config.softmax_selfnorm_alpha = softmax_selfnorm_alpha

Q
qijun 已提交
1604

Z
zhangjinchao01 已提交
1605 1606
@config_layer('fc')
class FCLayer(LayerBase):
T
tensor-tang 已提交
1607 1608
    layer_type = 'fc'

L
lianxiaochen 已提交
1609 1610 1611 1612 1613 1614 1615
    def __init__(self,
                 name,
                 size,
                 inputs,
                 bias=True,
                 error_clipping_threshold=None,
                 **xargs):
T
tensor-tang 已提交
1616
        use_mkldnn = bool(int(g_command_config_args.get("use_mkldnn", 0)))
1617 1618
        use_mkldnn_wgt = bool(
            int(g_command_config_args.get("use_mkldnn_wgt", 0)))
T
tensor-tang 已提交
1619 1620 1621 1622 1623 1624 1625
        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
1626 1627 1628
        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
            psize = self.config.size * input_layer.size
T
tensor-tang 已提交
1629
            dims = [input_layer.size, self.config.size]
Z
zhangjinchao01 已提交
1630 1631
            format = self.inputs[input_index].format
            sparse = format == "csr" or format == "csc"
T
tensor-tang 已提交
1632 1633 1634
            if use_mkldnn:
                config_assert(not sparse,
                              "MkldnnFCLayer do not support sparse format yet")
T
tensor-tang 已提交
1635 1636
                if use_mkldnn_wgt:
                    dims = [self.config.size, input_layer.size]
Z
zhangjinchao01 已提交
1637 1638
            if sparse:
                psize = self.inputs[input_index].nnz
1639 1640
            else:
                sparse = None
Z
zhangjinchao01 已提交
1641

Q
qijun 已提交
1642 1643
            self.create_input_parameter(input_index, psize, dims, sparse,
                                        format)
Z
zhangjinchao01 已提交
1644
        self.create_bias_parameter(bias, self.config.size)
L
lianxiaochen 已提交
1645 1646
        if error_clipping_threshold is not None:
            self.config.error_clipping_threshold = error_clipping_threshold
Z
zhangjinchao01 已提交
1647

Q
qijun 已提交
1648

T
tensor-tang 已提交
1649 1650 1651 1652 1653
@config_layer('mkldnn_fc')
class MkldnnFcLayer(FCLayer):
    layer_type = 'mkldnn_fc'


Z
zhangjinchao01 已提交
1654 1655
@config_layer('selective_fc')
class SelectiveFCLayer(LayerBase):
Q
qijun 已提交
1656 1657 1658 1659 1660 1661 1662 1663 1664 1665
    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
1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685
        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
1686 1687
                          ("if indices of selected columns are not specified, "
                           "selective_fc Layer has at least two inputs"))
Z
zhangjinchao01 已提交
1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699
            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
1700 1701
            self.create_input_parameter(input_index, psize, dims, sparse,
                                        format)
Z
zhangjinchao01 已提交
1702 1703
        self.create_bias_parameter(bias, self.config.size)

Q
qijun 已提交
1704

1705 1706
@config_layer('print')
class PrintLayer(LayerBase):
1707
    def __init__(self, name, inputs, format=None):
1708
        super(PrintLayer, self).__init__(name, 'print', 0, inputs)
1709 1710 1711 1712 1713 1714
        if format is None:
            format = "\n".join([
                "layer=" + input.input_layer_name + " %s"
                for input in self.inputs
            ])
        self.config.user_arg = format
1715

Q
qijun 已提交
1716

Y
yuan 已提交
1717 1718
@config_layer('priorbox')
class PriorBoxLayer(LayerBase):
G
gaoyuan 已提交
1719 1720
    def __init__(self, name, inputs, size, min_size, max_size, aspect_ratio,
                 variance):
Y
yuan 已提交
1721
        super(PriorBoxLayer, self).__init__(name, 'priorbox', 0, inputs)
G
gaoyuan 已提交
1722
        config_assert(len(inputs) == 2, 'PriorBoxLayer must have 2 inputs')
G
gaoyuan 已提交
1723 1724 1725 1726 1727 1728 1729
        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
1730
        config_assert(len(variance) == 4, 'The variance must have 4 inputs')
Y
yuan 已提交
1731 1732 1733 1734 1735 1736
        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
1737

1738 1739 1740
@config_layer('multibox_loss')
class MultiBoxLossLayer(LayerBase):
    def __init__(self, name, inputs, input_num, num_classes, overlap_threshold,
1741
                 neg_pos_ratio, neg_overlap, background_id, **xargs):
1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762
        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,
1763
                 background_id, **xargs):
1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783
        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
1784 1785
@config_layer('data')
class DataLayer(LayerBase):
L
Luo Tao 已提交
1786
    def __init__(self, name, size, height=None, width=None, device=None):
Q
qijun 已提交
1787 1788
        super(DataLayer, self).__init__(
            name, 'data', size, inputs=[], device=device)
L
Luo Tao 已提交
1789 1790
        if height and width:
            self.set_layer_height_width(height, width)
Q
qijun 已提交
1791

Z
zhangjinchao01 已提交
1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818

'''
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
1819 1820


Z
zhangjinchao01 已提交
1821 1822
@config_layer('data_norm')
class DataNormLayer(LayerBase):
Q
qijun 已提交
1823
    def __init__(self, name, inputs, data_norm_strategy="z-score", device=None):
Z
zhangjinchao01 已提交
1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834
        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
1835

Z
zhangjinchao01 已提交
1836 1837 1838
@config_layer('prelu')
class ParameterReluLayer(LayerBase):
    layer_type = 'prelu'
Q
qijun 已提交
1839 1840

    def __init__(self, name, inputs, partial_sum=1, **args):
Z
zhangjinchao01 已提交
1841 1842 1843
        super(ParameterReluLayer, self).__init__(
            name, self.layer_type, 0, inputs=inputs, **args)
        input_layer = self.get_input_layer(0)
1844 1845 1846
        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
1847 1848 1849
        self.set_layer_size(input_layer.size)
        self.create_input_parameter(0, input_layer.size / partial_sum)

Q
qijun 已提交
1850

Z
zhangjinchao01 已提交
1851 1852 1853
@config_layer('conv')
class ConvLayerBase(LayerBase):
    layer_type = 'conv'
Q
qijun 已提交
1854 1855 1856 1857 1858 1859 1860 1861

    def __init__(self,
                 name,
                 inputs=[],
                 bias=True,
                 num_filters=None,
                 shared_biases=False,
                 **xargs):
Z
zhangjinchao01 已提交
1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877
        super(ConvLayerBase, self).__init__(
            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))

        # Automatically select cudnn_type for GPU and exconv for CPU
        # if set type=conv, but still reserve the way user specify
        # exconv or cudnn_conv manually.
        if self.layer_type == "cudnn_conv":
            config_assert(use_gpu, "cudnn_conv only support GPU")

        if (use_gpu == 1 and self.layer_type != "exconv" and
Q
qijun 已提交
1878
            (parallel_nn == 0 or self.config.device > -1)):
Z
zhangjinchao01 已提交
1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890
            self.layer_type = "cudnn_conv"
        else:
            self.layer_type = "exconv"
        # 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
1891 1892
            parse_conv(self.inputs[input_index].conv, input_layer.name,
                       conv_conf, num_filters)
Z
zhangjinchao01 已提交
1893 1894
            psize = self.calc_parameter_size(conv_conf)
            self.create_input_parameter(input_index, psize)
L
Luo Tao 已提交
1895 1896
            self.set_cnn_layer(name, conv_conf.output_y, conv_conf.output_x,
                               self.config.num_filters)
Z
zhangjinchao01 已提交
1897 1898 1899 1900 1901 1902 1903 1904 1905 1906

        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)

Q
qijun 已提交
1907

Z
zhangjinchao01 已提交
1908 1909 1910 1911
@config_layer('exconv')
class ConvLayer(ConvLayerBase):
    layer_type = 'exconv'

Q
qijun 已提交
1912

Z
zhangjinchao01 已提交
1913 1914 1915 1916
@config_layer('cudnn_conv')
class ConvLayer(ConvLayerBase):
    layer_type = 'cudnn_conv'

1917 1918 1919 1920

@config_layer('convt')
class ConvTransLayerBase(LayerBase):
    layer_type = 'convt'
Q
qijun 已提交
1921 1922 1923 1924 1925 1926 1927 1928

    def __init__(self,
                 name,
                 inputs=[],
                 bias=True,
                 num_filters=None,
                 shared_biases=False,
                 **xargs):
1929
        super(ConvTransLayerBase, self).__init__(
1930 1931 1932 1933 1934 1935 1936 1937
            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))

1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948
        # 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"
1949 1950 1951 1952 1953 1954 1955 1956
        # 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)
1957
            parse_conv(
1958 1959
                self.inputs[input_index].conv,
                input_layer.name,
1960
                self.config.inputs[input_index].conv_conf,
1961
                num_filters,
1962
                trans=True)
1963 1964 1965
            conv_conf = self.config.inputs[input_index].conv_conf
            psize = self.calc_parameter_size(conv_conf)
            self.create_input_parameter(input_index, psize)
1966 1967
            self.set_cnn_layer(name, conv_conf.img_size_y, conv_conf.img_size,
                               self.config.num_filters)
1968 1969 1970 1971 1972 1973 1974

        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):
1975
        return conv_conf.channels * conv_conf.filter_channels \
1976 1977
                    * (conv_conf.filter_size * conv_conf.filter_size_y)

Q
qijun 已提交
1978

1979 1980 1981 1982
@config_layer('exconvt')
class ConvTransLayer(ConvTransLayerBase):
    layer_type = 'exconvt'

Q
qijun 已提交
1983

1984 1985 1986 1987 1988
@config_layer('cudnn_convt')
class ConvTransLayer(ConvTransLayerBase):
    layer_type = 'cudnn_convt'


Z
zhangjinchao01 已提交
1989 1990
@config_layer('norm')
class NormLayer(LayerBase):
1991 1992
    def __init__(self, name, inputs, **xargs):
        super(NormLayer, self).__init__(name, 'norm', 0, inputs=inputs, **xargs)
Z
zhangjinchao01 已提交
1993 1994 1995
        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
1996 1997 1998 1999
            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)
2000 2001 2002
            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
2003

Z
zhangjinchao01 已提交
2004 2005 2006

@config_layer('pool')
class PoolLayer(LayerBase):
2007 2008
    def __init__(self, name, inputs, ceil_mode=True, **xargs):
        super(PoolLayer, self).__init__(name, 'pool', 0, inputs=inputs, **xargs)
Z
zhangjinchao01 已提交
2009 2010 2011
        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
2012
            parse_pool(self.inputs[input_index].pool, input_layer.name,
2013
                       pool_conf, ceil_mode)
L
Luo Tao 已提交
2014 2015
            self.set_cnn_layer(name, pool_conf.output_y, pool_conf.output_x,
                               pool_conf.channels)
Q
qijun 已提交
2016

Z
zhangjinchao01 已提交
2017

Q
qijun 已提交
2018 2019
@config_layer('spp')
class SpatialPyramidPoolLayer(LayerBase):
2020
    def __init__(self, name, inputs, **xargs):
Q
qijun 已提交
2021
        super(SpatialPyramidPoolLayer, self).__init__(
2022
            name, 'spp', 0, inputs=inputs, **xargs)
Q
qijun 已提交
2023 2024 2025
        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
2026 2027 2028
            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
2029

Q
qijun 已提交
2030

D
dangqingqing 已提交
2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049
@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


2050 2051
@config_layer('crop')
class CropLayer(LayerBase):
2052
    def __init__(self, name, inputs, axis, offset, shape, **xargs):
2053
        super(CropLayer, self).__init__(name, 'crop', 0, inputs=inputs, **xargs)
2054 2055 2056
        self.config.axis = axis
        self.config.offset.extend(offset)
        self.config.shape.extend(shape)
2057 2058 2059 2060 2061 2062 2063 2064 2065 2066

        # 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
2067 2068 2069
@config_layer('batch_norm')
class BatchNormLayer(LayerBase):
    layer_type = 'batch_norm'
Q
qijun 已提交
2070 2071 2072 2073 2074 2075 2076 2077 2078

    def __init__(self,
                 name,
                 inputs,
                 bias=True,
                 use_global_stats=True,
                 moving_average_fraction=0.9,
                 batch_norm_type=None,
                 **xargs):
Z
zhangjinchao01 已提交
2079 2080 2081 2082
        if inputs is None:
            inputs = []
        elif not isinstance(inputs, list):
            inputs = [inputs]
Q
qijun 已提交
2083 2084
        config_assert(
            len(inputs) == 1, "BatchNormLayer must have one and only one input")
Z
zhangjinchao01 已提交
2085 2086 2087 2088 2089 2090 2091 2092
        # 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
2093 2094 2095 2096 2097 2098
            inputs.append(
                Input(
                    inputs[0].input_layer_name,
                    initial_std=0.0,
                    initial_mean=0.0,
                    is_static=True,
2099
                    is_shared=is_shared,
D
dangqingqing 已提交
2100
                    make_layer_name_in_submodel=False, ))
Z
zhangjinchao01 已提交
2101 2102 2103 2104 2105 2106

        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 \
2107
                ((not parallel_nn) or self.config.device > -1)
Z
zhangjinchao01 已提交
2108
        self.layer_type = "cudnn_batch_norm" if use_cudnn else "batch_norm"
Q
qijun 已提交
2109
        super(BatchNormLayer, self).__init__(
X
xuwei06 已提交
2110
            name, self.layer_type, 0, inputs=inputs, **xargs)
Z
zhangjinchao01 已提交
2111 2112 2113 2114 2115 2116

        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
2117
        input_layer = self.get_input_layer(0)
Z
zhangjinchao01 已提交
2118
        image_conf = self.config.inputs[0].image_conf
L
Luo Tao 已提交
2119
        parse_image(self.inputs[0].image, input_layer.name, image_conf)
2120

2121 2122
        # Only pass the width and height of input to batch_norm layer
        # when either of it is non-zero.
2123 2124
        if input_layer.width != 0 or input_layer.height != 0:
            self.set_cnn_layer(name, image_conf.img_size_y, image_conf.img_size,
D
dangqingqing 已提交
2125
                               image_conf.channels, False)
2126 2127
        else:
            self.set_layer_size(input_layer.size)
Z
zhangjinchao01 已提交
2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139

        psize = self.calc_parameter_size(image_conf)
        dims = [1, psize]
        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)

    def calc_parameter_size(self, image_conf):
        return image_conf.channels

Q
qijun 已提交
2140

Z
zhangjinchao01 已提交
2141 2142
@config_layer('trans')
class TransLayer(LayerBase):
2143
    def __init__(self, name, inputs, **xargs):
Q
qijun 已提交
2144
        super(TransLayer, self).__init__(
2145
            name, 'trans', 0, inputs=inputs, **xargs)
Q
qijun 已提交
2146 2147 2148
        config_assert(
            len(self.inputs) == 1,
            'TransLayer must have one and only one input')
Z
zhangjinchao01 已提交
2149 2150
        self.set_layer_size(self.get_input_layer(0).size)

Q
qijun 已提交
2151

Z
zhangjinchao01 已提交
2152 2153
@config_layer('resize')
class ResizeLayer(LayerBase):
2154
    def __init__(self, name, size, inputs, **xargs):
Q
qijun 已提交
2155
        super(ResizeLayer, self).__init__(
2156
            name, 'resize', size=size, inputs=inputs, **xargs)
Q
qijun 已提交
2157 2158 2159 2160
        config_assert(
            len(self.inputs) == 1,
            'ResizeLayer must have one and only one input')

Z
zhangjinchao01 已提交
2161

2162 2163
@config_layer('rotate')
class RotateLayer(LayerBase):
H
Haonan 已提交
2164
    def __init__(self, name, inputs, height, width, device=None):
2165 2166 2167 2168 2169
        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
2170
        self.set_layer_height_width(height, width)
2171 2172 2173
        self.set_layer_size(self.get_input_layer(0).size)


Z
zhangjinchao01 已提交
2174 2175
@config_layer('blockexpand')
class BlockExpandLayer(LayerBase):
2176
    def __init__(self, name, inputs, **xargs):
Q
qijun 已提交
2177
        super(BlockExpandLayer, self).__init__(
2178
            name, 'blockexpand', 0, inputs=inputs, **xargs)
Z
zhangjinchao01 已提交
2179 2180
        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
Q
qijun 已提交
2181 2182
            parse_block_expand(
                self.inputs[input_index].block_expand, input_layer.name,
Z
zhangjinchao01 已提交
2183
                self.config.inputs[input_index].block_expand_conf)
Q
qijun 已提交
2184 2185 2186 2187 2188 2189
            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
2190

2191 2192
@config_layer('maxout')
class MaxOutLayer(LayerBase):
Q
qijun 已提交
2193 2194 2195
    def __init__(self, name, inputs, **xargs):
        super(MaxOutLayer, self).__init__(
            name, 'maxout', 0, inputs=inputs, **xargs)
2196 2197
        input_layer = self.get_input_layer(0)
        maxout_conf = self.config.inputs[0].maxout_conf
L
Luo Tao 已提交
2198
        parse_maxout(self.inputs[0].maxout, input_layer.name, maxout_conf)
L
Luo Tao 已提交
2199 2200 2201
        out_channels = maxout_conf.image_conf.channels / maxout_conf.groups
        self.set_cnn_layer(name, g_layer_map[input_layer.name].height,
                           g_layer_map[input_layer.name].width, out_channels)
Q
qijun 已提交
2202

2203

D
dangqingqing 已提交
2204 2205 2206 2207
@config_layer('row_conv')
class RowConvLayer(LayerBase):
    def __init__(self, name, inputs, context_length, **xargs):
        super(RowConvLayer, self).__init__(
2208
            name, 'row_conv', 0, inputs=inputs, **xargs)
D
dangqingqing 已提交
2209 2210
        config_assert(
            len(self.inputs) == 1,
2211
            'row convolution layer must have one and only one input.')
D
dangqingqing 已提交
2212 2213 2214 2215 2216 2217 2218 2219 2220
        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
2221 2222
@config_layer('clip')
class ClipLayer(LayerBase):
2223 2224
    def __init__(self, name, inputs, min, max, **xargs):
        super(ClipLayer, self).__init__(name, 'clip', 0, inputs=inputs, **xargs)
G
guosheng 已提交
2225 2226
        config_assert(
            len(self.inputs) == 1,
2227 2228
            'ClipLayer must have one and only one input.')
        config_assert(min < max, 'min must be less than max.')
G
guosheng 已提交
2229 2230
        input_layer = self.get_input_layer(0)
        self.set_layer_size(input_layer.size)
2231 2232
        self.config.inputs[0].clip_conf.min = min
        self.config.inputs[0].clip_conf.max = max
G
guosheng 已提交
2233 2234


G
guosheng 已提交
2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248
@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
2249 2250 2251 2252
# key: cost type
# value: cost class
g_cost_map = {}

Q
qijun 已提交
2253

Z
zhangjinchao01 已提交
2254 2255 2256
# 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
2257 2258
        super(type(cls), cls).__init__(
            name, cost_type, 1, inputs, device=device, coeff=coeff)
Z
zhangjinchao01 已提交
2259

Q
qijun 已提交
2260
    cls = type(class_name, (LayerBase, ), dict(__init__=init))
Z
zhangjinchao01 已提交
2261 2262 2263
    global g_cost_map
    g_cost_map[cost_type] = cls

Q
qijun 已提交
2264

Z
zhangjinchao01 已提交
2265 2266 2267 2268 2269 2270 2271 2272
define_cost('MultiClassCrossEntropy', 'multi-class-cross-entropy')
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')
define_cost('HuberTwoClass', 'huber')
X
xuwei06 已提交
2273
define_cost('SumCost', 'sum_cost')
D
dangqingqing 已提交
2274
define_cost('SmoothL1Cost', 'smooth_l1')
Z
zhangjinchao01 已提交
2275

Q
qijun 已提交
2276

Z
zhangjinchao01 已提交
2277 2278
@config_layer('hsigmoid')
class HierarchicalSigmoidLayer(LayerBase):
Q
qijun 已提交
2279
    def __init__(self, name, num_classes, inputs, device=None, bias=True):
Z
zhangjinchao01 已提交
2280 2281
        super(HierarchicalSigmoidLayer, self).__init__(
            name, 'hsigmoid', 1, inputs=inputs, device=device)
Q
qijun 已提交
2282 2283 2284
        config_assert(
            len(self.inputs) >= 2,
            'HierarchicalSigmoidLayer must have at least 2 inputs')
Z
zhangjinchao01 已提交
2285 2286 2287 2288 2289 2290 2291 2292
        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
2293

Z
zhangjinchao01 已提交
2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317
'''
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
2318 2319


Z
zhangjinchao01 已提交
2320 2321
@config_layer('lambda_cost')
class LambdaCost(LayerBase):
Q
qijun 已提交
2322
    def __init__(self, name, inputs, NDCG_num=5, max_sort_size=-1, device=None):
Z
zhangjinchao01 已提交
2323 2324
        super(LambdaCost, self).__init__(
            name, 'lambda_cost', 1, inputs=inputs, device=device)
Q
qijun 已提交
2325
        config_assert(len(self.inputs) == 2, 'lambdaCost must have 2 inputs')
Z
zhangjinchao01 已提交
2326 2327
        self.config.NDCG_num = NDCG_num
        if max_sort_size != -1:
Q
qijun 已提交
2328 2329 2330
            config_assert(
                NDCG_num <= max_sort_size,
                'NDCG_num must be less than or equal to max_sort_size')
Z
zhangjinchao01 已提交
2331 2332
        self.config.max_sort_size = max_sort_size

Q
qijun 已提交
2333

Z
zhangjinchao01 已提交
2334 2335
@config_layer('nce')
class NCELayer(LayerBase):
Q
qijun 已提交
2336 2337 2338 2339 2340 2341 2342 2343
    def __init__(self,
                 name,
                 num_classes,
                 inputs,
                 num_neg_samples=10,
                 neg_sampling_dist=None,
                 bias=True,
                 **xargs):
Z
zhangjinchao01 已提交
2344
        super(NCELayer, self).__init__(name, 'nce', 1, inputs=inputs, **xargs)
Q
qijun 已提交
2345 2346
        config_assert(
            len(self.inputs) >= 2, 'NCELayer must have at least 2 inputs')
Z
zhangjinchao01 已提交
2347 2348
        self.config.num_classes = num_classes
        if neg_sampling_dist is not None:
Q
qijun 已提交
2349 2350 2351 2352
            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
2353
            s = sum(neg_sampling_dist)
Q
qijun 已提交
2354 2355 2356
            config_assert(
                abs(s - 1) < 1e-5,
                'The sum of neg_sampling_dist (%s) is not 1' % s)
Z
zhangjinchao01 已提交
2357 2358 2359 2360 2361

            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
2362
        input_layer = self.get_input_layer(num_real_inputs)
Z
zhangjinchao01 已提交
2363 2364 2365 2366
        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
2367 2368
        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
2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381
            # 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
2382
    def __init__(self, name, inputs, bias=True, **xargs):
Z
zhangjinchao01 已提交
2383 2384
        super(AddToLayer, self).__init__(
            name, 'addto', 0, inputs=inputs, **xargs)
Q
qijun 已提交
2385
        config_assert(len(inputs) > 0, 'inputs cannot be empty for AddToLayer')
Z
zhangjinchao01 已提交
2386 2387 2388 2389 2390
        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
2391

Z
zhangjinchao01 已提交
2392 2393
@config_layer('agent')
class AgentLayer(LayerBase):
Q
qijun 已提交
2394 2395 2396 2397
    def __init__(self, name, size, device=None):
        super(AgentLayer, self).__init__(
            name, 'agent', size, inputs=[], device=device)

Z
zhangjinchao01 已提交
2398 2399 2400

@config_layer('gather_agent')
class GatherAgentLayer(LayerBase):
Q
qijun 已提交
2401
    def __init__(self, name, size, device=None):
Z
zhangjinchao01 已提交
2402 2403 2404
        super(GatherAgentLayer, self).__init__(
            name, 'gather_agent', size, inputs=[], device=device)

Q
qijun 已提交
2405

Z
zhangjinchao01 已提交
2406 2407
@config_layer('scatter_agent')
class ScatterAgentLayer(LayerBase):
Q
qijun 已提交
2408
    def __init__(self, name, size, device=None):
Z
zhangjinchao01 已提交
2409 2410 2411
        super(ScatterAgentLayer, self).__init__(
            name, 'scatter_agent', size, inputs=[], device=device)

Q
qijun 已提交
2412

Z
zhangjinchao01 已提交
2413 2414
@config_layer('multiplex')
class MultiplexLayer(LayerBase):
Q
qijun 已提交
2415 2416 2417 2418 2419
    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
2420
        for i in range(1, len(inputs)):
Q
qijun 已提交
2421 2422 2423 2424 2425
            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
2426 2427

@config_func
2428 2429 2430 2431
def Link(name, has_subseq=False):
    """
    Still keeping has_subseq for backward compatibility
    """
Z
zhangjinchao01 已提交
2432 2433 2434 2435
    link_config = LinkConfig()
    link_config.link_name = name
    return link_config

Q
qijun 已提交
2436

Z
zhangjinchao01 已提交
2437 2438
# memory for recurrent layer group.
# *name* and *size* are actual layer's name and size.
2439 2440 2441 2442
# 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
2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453
# 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
2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465
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
2466
    agent_layer = AgentLayer(agent_name, size)
Z
zhangjinchao01 已提交
2467
    config_assert(g_current_submodel.is_recurrent_layer_group,
Q
qijun 已提交
2468
                  'Memory should be used in recurrent layer group only')
Z
zhangjinchao01 已提交
2469
    memory = g_current_submodel.memories.add()
2470 2471
    if name is not None:
        memory.layer_name = MakeLayerNameInSubmodel(name)
Z
zhangjinchao01 已提交
2472
    memory.link_name = MakeLayerNameInSubmodel(agent_name)
Q
qijun 已提交
2473
    options = sum((boot_layer is not None, bool(boot_bias),
Z
zhangjinchao01 已提交
2474
                   boot_with_const_id is not None))
Q
qijun 已提交
2475 2476 2477 2478
    config_assert(
        options <= 1,
        'take one option at most from boot_layer, boot_bias, or boot_with_const_id'
    )
Z
zhangjinchao01 已提交
2479 2480 2481
    if boot_layer is not None:
        boot_layer = MakeLayerNameInParentSubmodel(boot_layer)
        config_assert(boot_layer in g_layer_map,
Q
qijun 已提交
2482 2483
                      'boot_layer "%s" does not correspond to a layer name' %
                      boot_layer)
Z
zhangjinchao01 已提交
2484 2485 2486
        memory.boot_layer_name = boot_layer
    elif boot_bias:
        memory.boot_bias_parameter_name = agent_layer.create_bias_parameter(
Q
qijun 已提交
2487
            boot_bias, size, for_self=False)
Z
zhangjinchao01 已提交
2488 2489 2490 2491 2492
        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
2493

2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504
@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
2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515
# 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
2516 2517 2518 2519
        eos_layer_name="eos_check",
        num_results_per_sample=1,
        beam_size=1,
        log_prob=None, ):
Z
zhangjinchao01 已提交
2520 2521 2522 2523 2524 2525 2526 2527 2528
    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
2529

Z
zhangjinchao01 已提交
2530 2531
@config_layer('expand')
class ExpandLayer(LayerBase):
2532
    def __init__(self, name, inputs, trans_type='non-seq', bias=False, **xargs):
Q
qijun 已提交
2533
        super(ExpandLayer, self).__init__(
2534
            name, 'expand', 0, inputs=inputs, **xargs)
Q
qijun 已提交
2535 2536 2537 2538 2539 2540 2541 2542
        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
2543 2544 2545

@config_layer('featmap_expand')
class FeatMapExpandLayer(LayerBase):
X
xuwei06 已提交
2546 2547 2548 2549 2550
    def __init__(self,
                 name,
                 inputs,
                 num_filters=None,
                 as_row_vector=True,
X
xuwei06 已提交
2551 2552
                 bias=False,
                 **xargs):
Q
qijun 已提交
2553
        super(FeatMapExpandLayer, self).__init__(
X
xuwei06 已提交
2554
            name, 'featmap_expand', 0, inputs=inputs, **xargs)
Q
qijun 已提交
2555 2556 2557
        config_assert(
            len(self.inputs) == 1, 'ExpandLayer takes 1 and only 1 inputs')
        if num_filters is not None:
Z
zhangjinchao01 已提交
2558
            self.config.num_filters = num_filters
Q
qijun 已提交
2559
        else:
Z
zhangjinchao01 已提交
2560
            logger.fatal("FeatMapExpandLayer must specify num_filters.")
X
xuwei06 已提交
2561 2562
        if not as_row_vector:
            self.config.user_arg = "as_col_vec"
Q
qijun 已提交
2563
        self.set_layer_size(self.get_input_layer(0).size * num_filters)
Z
zhangjinchao01 已提交
2564 2565 2566 2567


@config_layer('max')
class MaxLayer(LayerBase):
Q
qijun 已提交
2568 2569 2570 2571 2572
    def __init__(self,
                 name,
                 inputs,
                 trans_type='non-seq',
                 bias=False,
2573
                 output_max_index=None,
2574
                 stride=-1,
2575
                 **xargs):
2576
        super(MaxLayer, self).__init__(name, 'max', 0, inputs=inputs, **xargs)
Z
zhangjinchao01 已提交
2577
        config_assert(len(self.inputs) == 1, 'MaxLayer must have 1 input')
2578 2579
        if trans_type == 'seq':
            config_assert(stride == -1, 'subseq does not support stride window')
Q
qijun 已提交
2580
        self.config.trans_type = trans_type
2581
        self.config.seq_pool_stride = stride
Z
zhangjinchao01 已提交
2582 2583 2584 2585
        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)
2586 2587
        if output_max_index is not None:
            self.config.output_max_index = output_max_index
Z
zhangjinchao01 已提交
2588 2589 2590 2591


@config_layer('maxid')
class MaxIdLayer(LayerBase):
Q
qijun 已提交
2592
    def __init__(self, name, inputs, beam_size=None, device=None):
Z
zhangjinchao01 已提交
2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609
        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
2610
    def __init__(self, name, inputs, eos_id, device=None):
Z
zhangjinchao01 已提交
2611 2612 2613
        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
2614
        self.set_layer_size(2)  # boolean output
Z
zhangjinchao01 已提交
2615 2616
        self.config.eos_id = eos_id

Q
qijun 已提交
2617

Z
zhangjinchao01 已提交
2618 2619
@config_layer('seqlastins')
class SequenceLastInstanceLayer(LayerBase):
Q
qijun 已提交
2620 2621 2622 2623
    def __init__(self,
                 name,
                 inputs,
                 trans_type='non-seq',
2624
                 bias=False,
2625
                 stride=-1,
2626
                 **xargs):
Q
qijun 已提交
2627
        super(SequenceLastInstanceLayer, self).__init__(
X
xuwei06 已提交
2628
            name, 'seqlastins', 0, inputs=inputs, **xargs)
Q
qijun 已提交
2629 2630
        config_assert(
            len(inputs) == 1, 'SequenceLastInstanceLayer must have 1 input')
2631
        if trans_type == 'seq':
L
Luo Tao 已提交
2632
            config_assert(stride == -1, 'subseq does not support stride window')
Q
qijun 已提交
2633
        self.config.trans_type = trans_type
2634 2635
        self.config.seq_pool_stride = stride
        self.set_layer_size(self.get_input_layer(0).size)
Z
zhangjinchao01 已提交
2636 2637
        self.create_bias_parameter(bias, self.config.size)

Q
qijun 已提交
2638

Z
zhangjinchao01 已提交
2639 2640
@config_layer('seqfirstins')
class SequenceFirstInstanceLayer(SequenceLastInstanceLayer):
2641 2642 2643 2644 2645
    def __init__(self,
                 name,
                 inputs,
                 trans_type='non-seq',
                 bias=False,
2646
                 stride=-1,
2647
                 **xargs):
Q
qijun 已提交
2648
        super(SequenceFirstInstanceLayer, self).__init__(
2649 2650 2651 2652 2653 2654
            name,
            inputs=inputs,
            trans_type=trans_type,
            bias=bias,
            stride=stride,
            **xargs)
Z
zhangjinchao01 已提交
2655 2656
        self.config.select_first = True

Q
qijun 已提交
2657

Z
zhangjinchao01 已提交
2658 2659
@config_layer('seqconcat')
class SequenceConcatLayer(LayerBase):
X
xuwei06 已提交
2660
    def __init__(self, name, inputs, bias=False, **xargs):
Q
qijun 已提交
2661
        super(SequenceConcatLayer, self).__init__(
X
xuwei06 已提交
2662
            name, 'seqconcat', 0, inputs=inputs, **xargs)
Q
qijun 已提交
2663 2664
        config_assert(
            len(inputs) == 2, 'SequenceConcatLayer must have 2 inputs')
Z
zhangjinchao01 已提交
2665 2666 2667 2668 2669
        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
2670

Z
zhangjinchao01 已提交
2671 2672
@config_layer('seqreshape')
class SequenceReshapeLayer(LayerBase):
X
xuwei06 已提交
2673
    def __init__(self, name, size, inputs, bias=False, **xargs):
Q
qijun 已提交
2674
        super(SequenceReshapeLayer, self).__init__(
X
xuwei06 已提交
2675
            name, 'seqreshape', size, inputs=inputs, **xargs)
Q
qijun 已提交
2676 2677
        config_assert(
            len(inputs) == 1, 'SequenceReshapeLayer must have 1 inputs')
Z
zhangjinchao01 已提交
2678 2679 2680
        self.set_layer_size(size)
        self.create_bias_parameter(bias, size)

Q
qijun 已提交
2681

Z
zhangjinchao01 已提交
2682 2683
@config_layer('subseq')
class SubSequenceLayer(LayerBase):
X
xuwei06 已提交
2684
    def __init__(self, name, inputs, bias=False, **xargs):
Q
qijun 已提交
2685
        super(SubSequenceLayer, self).__init__(
X
xuwei06 已提交
2686
            name, 'subseq', 0, inputs=inputs, **xargs)
Z
zhangjinchao01 已提交
2687 2688 2689 2690 2691 2692
        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
2693

C
caoying03 已提交
2694 2695
@config_layer('sub_nested_seq')
class SubNestedSequenceLayer(LayerBase):
2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707
    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
2708
        super(SubNestedSequenceLayer, self).__init__(
2709 2710 2711 2712 2713
            name,
            'sub_nested_seq',
            0,
            inputs=[inputs, selected_indices],
            **xargs)
C
caoying03 已提交
2714 2715 2716 2717 2718
        input_layer0 = self.get_input_layer(0)
        size = input_layer0.size
        self.set_layer_size(size)


Z
zhangjinchao01 已提交
2719 2720
@config_layer('out_prod')
class OuterProdLayer(LayerBase):
Q
qijun 已提交
2721 2722 2723
    def __init__(self, name, inputs, device=None):
        super(OuterProdLayer, self).__init__(
            name, 'out_prod', 0, inputs=inputs, device=device)
Z
zhangjinchao01 已提交
2724 2725 2726 2727 2728
        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
2729

Z
zhangjinchao01 已提交
2730 2731
@config_layer('power')
class PowerLayer(LayerBase):
Q
qijun 已提交
2732 2733 2734
    def __init__(self, name, inputs, device=None):
        super(PowerLayer, self).__init__(
            name, 'power', 0, inputs=inputs, device=device)
Z
zhangjinchao01 已提交
2735 2736 2737 2738
        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
2739 2740 2741
        config_assert(1 == input_layer0.size,
                      'The left input is the exponent and should be of size 1')

Z
zhangjinchao01 已提交
2742 2743 2744

@config_layer('slope_intercept')
class SlopeInterceptLayer(LayerBase):
Q
qijun 已提交
2745 2746 2747
    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
2748 2749 2750 2751 2752 2753
        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
2754

Z
zhangjinchao01 已提交
2755 2756
@config_layer('scaling')
class ScalingLayer(LayerBase):
Q
qijun 已提交
2757 2758 2759
    def __init__(self, name, inputs, device=None):
        super(ScalingLayer, self).__init__(
            name, 'scaling', 0, inputs=inputs, device=device)
Z
zhangjinchao01 已提交
2760 2761 2762 2763
        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
2764 2765 2766
        config_assert(1 == input_layer0.size,
                      'The left input should be of size 1')

Z
zhangjinchao01 已提交
2767 2768 2769

@config_layer('conv_shift')
class ConvShiftLayer(LayerBase):
Q
qijun 已提交
2770 2771 2772
    def __init__(self, name, inputs, device=None):
        super(ConvShiftLayer, self).__init__(
            name, 'conv_shift', 0, inputs=inputs, device=device)
Z
zhangjinchao01 已提交
2773 2774 2775 2776
        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
2777

Z
zhangjinchao01 已提交
2778 2779
@config_layer('convex_comb')
class ConvexCombinationLayer(LayerBase):
Q
qijun 已提交
2780
    def __init__(self, name, size, inputs, device=None):
Z
zhangjinchao01 已提交
2781
        super(ConvexCombinationLayer, self).__init__(
Q
qijun 已提交
2782 2783 2784
            name, 'convex_comb', size, inputs=inputs, device=device)
        config_assert(
            len(self.inputs) == 2, 'ConvexCombinationLayer must have 2 inputs')
2785 2786 2787
        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
2788 2789
        self.set_layer_size(size)

Q
qijun 已提交
2790

Z
zhangjinchao01 已提交
2791 2792
@config_layer('interpolation')
class InterpolationLayer(LayerBase):
Q
qijun 已提交
2793
    def __init__(self, name, inputs, device=None):
Z
zhangjinchao01 已提交
2794 2795
        super(InterpolationLayer, self).__init__(
            name, 'interpolation', 0, inputs=inputs, device=device)
Q
qijun 已提交
2796 2797
        config_assert(
            len(self.inputs) == 3, 'InterpolationLayer must have 3 inputs')
Z
zhangjinchao01 已提交
2798 2799 2800 2801 2802 2803 2804 2805
        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
2806

L
liaogang 已提交
2807 2808
@config_layer('bilinear_interp')
class BilinearInterpLayer(LayerBase):
Q
qijun 已提交
2809
    def __init__(self, name, inputs, **xargs):
L
liaogang 已提交
2810
        super(BilinearInterpLayer, self).__init__(
L
liaogang 已提交
2811
            name, 'bilinear_interp', 0, inputs=inputs, **xargs)
L
liaogang 已提交
2812
        input_layer = self.get_input_layer(0)
L
Luo Tao 已提交
2813 2814 2815 2816
        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
2817

L
liaogang 已提交
2818

Z
zhangjinchao01 已提交
2819 2820
@config_layer('sum_to_one_norm')
class SumToOneNormLayer(LayerBase):
Q
qijun 已提交
2821
    def __init__(self, name, inputs, device=None):
Z
zhangjinchao01 已提交
2822
        super(SumToOneNormLayer, self).__init__(
Q
qijun 已提交
2823 2824 2825
            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
2826 2827 2828
        input_layer0 = self.get_input_layer(0)
        self.set_layer_size(input_layer0.size)

Q
qijun 已提交
2829

G
guosheng 已提交
2830 2831
@config_layer('row_l2_norm')
class RowL2NormLayer(LayerBase):
2832
    def __init__(self, name, inputs, **xargs):
G
guosheng 已提交
2833
        super(RowL2NormLayer, self).__init__(
2834
            name, 'row_l2_norm', 0, inputs=inputs, **xargs)
G
guosheng 已提交
2835
        config_assert(len(self.inputs) == 1, 'RowL2NormLayer must have 1 input')
2836 2837
        input_layer = self.get_input_layer(0)
        self.set_layer_size(input_layer.size)
G
guosheng 已提交
2838 2839


Z
zhangjinchao01 已提交
2840 2841
@config_layer('cos_vm')
class CosSimVecMatLayer(LayerBase):
Q
qijun 已提交
2842
    def __init__(self, name, size, inputs, cos_scale=1.0, device=None):
Z
zhangjinchao01 已提交
2843
        super(CosSimVecMatLayer, self).__init__(
Q
qijun 已提交
2844
            name, 'cos_vm', size, inputs=inputs, device=device)
Z
zhangjinchao01 已提交
2845
        self.config.cos_scale = cos_scale
Q
qijun 已提交
2846 2847
        config_assert(
            len(self.inputs) == 2, 'CosSimVecMatLayer must have 2 inputs')
2848 2849 2850
        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
2851

Q
qijun 已提交
2852

Z
zhangjinchao01 已提交
2853 2854
@config_layer('sampling_id')
class SamplingIdLayer(LayerBase):
Q
qijun 已提交
2855
    def __init__(self, name, inputs, device=None):
Z
zhangjinchao01 已提交
2856 2857
        super(SamplingIdLayer, self).__init__(
            name, 'sampling_id', 0, inputs=inputs, device=device)
Q
qijun 已提交
2858 2859
        config_assert(
            len(self.inputs) == 1, 'SamplingIdLayer must have 1 input')
Z
zhangjinchao01 已提交
2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871
        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
2872 2873 2874 2875 2876
    def __init__(self,
                 name,
                 inputs,
                 average_strategy='average',
                 trans_type='non-seq',
2877
                 bias=False,
2878
                 stride=-1,
2879
                 **xargs):
Q
qijun 已提交
2880
        super(AverageLayer, self).__init__(
X
xuwei06 已提交
2881
            name, 'average', 0, inputs=inputs, **xargs)
Z
zhangjinchao01 已提交
2882
        self.config.average_strategy = average_strategy
2883 2884
        if trans_type == 'seq':
            config_assert(stride == -1, 'subseq does not support stride window')
Q
qijun 已提交
2885
        self.config.trans_type = trans_type
2886
        self.config.seq_pool_stride = stride
Z
zhangjinchao01 已提交
2887 2888 2889 2890 2891 2892
        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
2893

Z
zhangjinchao01 已提交
2894 2895
@config_layer('cos')
class CosSimLayer(LayerBase):
2896
    def __init__(self, name, inputs, cos_scale=1, device=None):
Z
zhangjinchao01 已提交
2897 2898 2899 2900 2901 2902
        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')
2903
        self.config.cos_scale = cos_scale
Z
zhangjinchao01 已提交
2904 2905 2906 2907


@config_layer('tensor')
class TensorLayer(LayerBase):
2908
    def __init__(self, name, size, inputs, bias=True, **xargs):
Q
qijun 已提交
2909
        super(TensorLayer, self).__init__(
2910
            name, 'tensor', size, inputs=inputs, **xargs)
Z
zhangjinchao01 已提交
2911 2912
        config_assert(len(self.inputs) == 2, 'TensorLayer must have 2 inputs')
        config_assert(size > 0, 'size must be positive')
Q
qijun 已提交
2913 2914
        config_assert(inputs[1].parameter_name == None,
                      'second parameter should be None.')
Z
zhangjinchao01 已提交
2915 2916 2917 2918 2919 2920 2921 2922 2923 2924
        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
2925
    def __init__(self, name, inputs, size=0, bias=True, **xargs):
Z
zhangjinchao01 已提交
2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942
        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)
2943
            if self.config.size == 0:
Z
zhangjinchao01 已提交
2944 2945 2946
                size = operator.calc_output_size(operator_conf.input_sizes)
                if size != 0:
                    self.set_layer_size(size)
2947
            else:
2948 2949
                sz = operator.calc_output_size(operator_conf.input_sizes)
                if sz != 0:
Q
qijun 已提交
2950 2951 2952 2953
                    config_assert(
                        sz == self.config.size,
                        "different inputs have different size: %s vs. %s" %
                        (sz, self.config.size))
Z
zhangjinchao01 已提交
2954 2955 2956 2957
        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
2958 2959 2960
                config_assert(
                    isinstance(input, Projection),
                    "input should be projection or operation")
2961
            if self.config.size == 0 and isinstance(input, Projection):
Z
zhangjinchao01 已提交
2962 2963 2964
                size = input.calc_output_size(input_layer)
                if size != 0:
                    self.set_layer_size(size)
2965
            elif isinstance(input, Projection):
Q
qijun 已提交
2966 2967 2968 2969 2970 2971
                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
2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982
        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
2983 2984
                input_config.proj_conf.name = gen_parameter_name(name,
                                                                 input_index)
Z
zhangjinchao01 已提交
2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995
                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)

2996 2997 2998 2999 3000 3001
        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
3002

3003 3004 3005
        if bias:
            self.config.bias_size = psize
            self.create_bias_parameter(bias, psize)
Z
zhangjinchao01 已提交
3006

Q
qijun 已提交
3007

Z
zhangjinchao01 已提交
3008 3009
# like MixedLayer, but no bias parameter
@config_func
Q
qijun 已提交
3010
def ExpressionLayer(name, inputs, **xargs):
Z
zhangjinchao01 已提交
3011 3012
    MixedLayer(name, inputs, bias=False, **xargs)

Q
qijun 已提交
3013

Z
zhangjinchao01 已提交
3014 3015
@config_layer('concat')
class ConcatenateLayer(LayerBase):
Q
qijun 已提交
3016
    def __init__(self, name, inputs, bias=False, **xargs):
Z
zhangjinchao01 已提交
3017
        config_assert(inputs, 'inputs cannot be empty')
3018
        config_assert(not bias, 'ConcatenateLayer cannot support bias.')
Z
zhangjinchao01 已提交
3019 3020 3021 3022 3023 3024
        super(ConcatenateLayer, self).__init__(
            name, 'concat', 0, inputs=inputs, **xargs)
        size = 0
        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
            input = self.inputs[input_index]
Q
qijun 已提交
3025
            if self.config.size == 0:
Z
zhangjinchao01 已提交
3026 3027 3028 3029
                size += input_layer.size

        self.set_layer_size(size)

Q
qijun 已提交
3030

Z
zhangjinchao01 已提交
3031 3032 3033
# like concat layer, but each input layer was processed by a Projection.
@config_layer('concat2')
class ConcatenateLayer2(LayerBase):
Q
qijun 已提交
3034
    def __init__(self, name, inputs, bias=False, **xargs):
Z
zhangjinchao01 已提交
3035 3036 3037
        config_assert(inputs, 'inputs cannot be empty')
        super(ConcatenateLayer2, self).__init__(
            name, 'concat2', 0, inputs=inputs, **xargs)
3038 3039

        if isinstance(self.inputs[0], ConvProjection):
Q
qijun 已提交
3040 3041 3042 3043 3044 3045
            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.")
3046

Z
zhangjinchao01 已提交
3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066
        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
3067
                                              input.proj_conf.output_size)
Z
zhangjinchao01 已提交
3068
            dims = input.calc_parameter_dims(input.proj_conf.input_size,
Q
qijun 已提交
3069
                                             input.proj_conf.output_size)
Z
zhangjinchao01 已提交
3070 3071
            self.create_input_parameter(input_index, psize, dims)

3072 3073 3074 3075 3076 3077 3078
        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()

3079 3080 3081
        if bias:
            self.config.bias_size = psize
            self.create_bias_parameter(bias, psize)
3082

Q
qijun 已提交
3083

Z
zhangjinchao01 已提交
3084 3085
@config_layer('recurrent')
class RecurrentLayer(LayerBase):
Q
qijun 已提交
3086
    def __init__(self, name, inputs, reversed=False, bias=True, **xargs):
Y
Yu Yang 已提交
3087 3088
        super(RecurrentLayer, self).__init__(name, 'recurrent', 0, inputs,
                                             **xargs)
Z
zhangjinchao01 已提交
3089 3090 3091 3092 3093 3094 3095 3096 3097
        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
3098

Z
zhangjinchao01 已提交
3099 3100
@config_layer('lstmemory')
class LstmLayer(LayerBase):
Q
qijun 已提交
3101 3102 3103 3104 3105 3106 3107 3108
    def __init__(self,
                 name,
                 inputs,
                 reversed=False,
                 active_gate_type="sigmoid",
                 active_state_type="sigmoid",
                 bias=True,
                 **xargs):
Z
zhangjinchao01 已提交
3109 3110 3111 3112 3113 3114 3115 3116
        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
3117
        self.config.active_gate_type = active_gate_type
Z
zhangjinchao01 已提交
3118 3119 3120 3121 3122
        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
3123

Z
zhangjinchao01 已提交
3124 3125
@config_layer('lstm_step')
class LstmStepLayer(LayerBase):
Q
qijun 已提交
3126 3127 3128 3129 3130 3131 3132 3133 3134 3135
    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
3136 3137 3138
        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
3139 3140 3141 3142 3143
        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
3144 3145 3146
        self.config.active_state_type = active_state_type
        self.create_bias_parameter(bias, size * 3)

Q
qijun 已提交
3147

Z
zhangjinchao01 已提交
3148 3149 3150
# get the specific output from the input layer.
@config_layer('get_output')
class GetOutputLayer(LayerBase):
Q
qijun 已提交
3151 3152 3153 3154
    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
3155 3156 3157 3158
        inputs = self.inputs[0]
        config_assert(inputs.input_layer_argument,
                      'input_layer_argument cannot be empty')

Q
qijun 已提交
3159

Z
zhangjinchao01 已提交
3160 3161
@config_layer('mdlstmemory')
class MDLstmLayer(LayerBase):
Q
qijun 已提交
3162 3163 3164 3165 3166 3167 3168 3169
    def __init__(self,
                 name,
                 inputs,
                 directions=True,
                 active_gate_type="sigmoid",
                 active_state_type="sigmoid",
                 bias=True,
                 **xargs):
Y
Yu Yang 已提交
3170 3171
        super(MDLstmLayer, self).__init__(name, 'mdlstmemory', 0, inputs,
                                          **xargs)
Z
zhangjinchao01 已提交
3172 3173 3174 3175
        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
3176 3177
        config_assert(input_layer.size % (3 + dim_num) == 0,
                      "size % (dim_num) should be 0!")
Q
qijun 已提交
3178
        size = input_layer.size / (3 + dim_num)
Z
zhangjinchao01 已提交
3179
        self.set_layer_size(size)
Q
qijun 已提交
3180
        self.config.active_gate_type = active_gate_type
Z
zhangjinchao01 已提交
3181 3182 3183
        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
3184 3185
        self.create_input_parameter(0, size * size * (3 + dim_num),
                                    [size, size, 3 + dim_num])
Z
zhangjinchao01 已提交
3186
        #bias includes 3 kinds of peephole, 3+dim_num+2+dim_num
Q
qijun 已提交
3187 3188
        self.create_bias_parameter(bias, size * (5 + 2 * dim_num))

Z
zhangjinchao01 已提交
3189 3190 3191

@config_layer('gated_recurrent')
class GatedRecurrentLayer(LayerBase):
Q
qijun 已提交
3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202
    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
3203 3204 3205 3206 3207 3208
        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
3209
        self.config.active_gate_type = active_gate_type
Z
zhangjinchao01 已提交
3210 3211 3212
        self.create_input_parameter(0, size * size * 3, [size, size * 3])
        self.create_bias_parameter(bias, size * 3)

Q
qijun 已提交
3213

Z
zhangjinchao01 已提交
3214 3215
@config_layer('gru_step')
class GruStepLayer(LayerBase):
Q
qijun 已提交
3216 3217 3218 3219 3220 3221 3222
    def __init__(self,
                 name,
                 size,
                 inputs,
                 active_gate_type="sigmoid",
                 bias=True,
                 **xargs):
Y
Yu Yang 已提交
3223 3224
        super(GruStepLayer, self).__init__(name, 'gru_step', size, inputs,
                                           **xargs)
Z
zhangjinchao01 已提交
3225 3226 3227
        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
3228 3229 3230 3231 3232
        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
3233
        self.create_input_parameter(0, size * size * 3, [size, size * 3])
Z
zhangjinchao01 已提交
3234 3235
        self.create_bias_parameter(bias, size * 3)

Q
qijun 已提交
3236

Z
zhangjinchao01 已提交
3237 3238 3239 3240 3241 3242 3243
'''
 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
3244 3245


Z
zhangjinchao01 已提交
3246 3247
@config_layer('crf')
class CRFLayer(LayerBase):
Q
qijun 已提交
3248
    def __init__(self, name, size, inputs, coeff=1.0, device=None):
Z
zhangjinchao01 已提交
3249
        super(CRFLayer, self).__init__(name, 'crf', size, inputs, device=device)
Q
qijun 已提交
3250 3251
        config_assert(2 <= len(self.inputs) <= 3,
                      'CRFLayer must have 2 or 3 inputs')
3252
        self.create_input_parameter(0, size * (size + 2), [size + 2, size])
Z
zhangjinchao01 已提交
3253 3254
        self.config.coeff = coeff

Q
qijun 已提交
3255

Z
zhangjinchao01 已提交
3256 3257 3258 3259 3260 3261 3262 3263
'''
 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
3264 3265


Z
zhangjinchao01 已提交
3266 3267
@config_layer('crf_decoding')
class CRFDecodingLayer(LayerBase):
Q
qijun 已提交
3268
    def __init__(self, name, size, inputs, device=None):
Z
zhangjinchao01 已提交
3269 3270 3271 3272 3273
        super(CRFDecodingLayer, self).__init__(
            name, 'crf_decoding', size, inputs, device=device)
        config_assert(
            len(self.inputs) <= 2,
            'CRFDecodingLayer cannot have more than 2 inputs')
3274
        self.create_input_parameter(0, size * (size + 2), [size + 2, size])
Z
zhangjinchao01 已提交
3275

Q
qijun 已提交
3276

Z
zhangjinchao01 已提交
3277 3278
@config_layer('ctc')
class CTCLayer(LayerBase):
Q
qijun 已提交
3279
    def __init__(self, name, size, inputs, norm_by_times=False, device=None):
Z
zhangjinchao01 已提交
3280 3281 3282 3283
        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
3284

3285 3286 3287 3288 3289 3290 3291 3292 3293 3294
@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


3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315
@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
3316 3317
@config_layer('recurrent_layer_group')
class RecurrentLayerGroup(LayerBase):
Q
qijun 已提交
3318
    def __init__(self, name, device=None):
Z
zhangjinchao01 已提交
3319 3320 3321 3322 3323 3324
        super(RecurrentLayerGroup, self).__init__(
            name, 'recurrent_layer_group', 0, inputs=[], device=device)


# Deprecated, use a new layer specific class instead
@config_func
Q
qijun 已提交
3325
def Layer(name, type, **xargs):
Z
zhangjinchao01 已提交
3326 3327 3328 3329
    layers = {}
    layers.update(g_cost_map)
    layers.update(g_layer_type_map)
    layer_func = layers.get(type)
Q
qijun 已提交
3330
    config_assert(layer_func, "layer type '%s' not supported." % type)
X
xuwei06 已提交
3331
    return layer_func(name, **xargs)
Z
zhangjinchao01 已提交
3332

Q
qijun 已提交
3333

Z
zhangjinchao01 已提交
3334
@config_func
Q
qijun 已提交
3335
def ParameterHook(type, **kwargs):
3336
    if type == 'pruning':
Z
zhangjinchao01 已提交
3337 3338
        hook = ParameterUpdaterHookConfig()
        hook.type = type
X
xzl 已提交
3339
        sparsity_ratio = kwargs.get('sparsity_ratio', None)
X
xzl 已提交
3340 3341
        if sparsity_ratio is not None:
            hook.sparsity_ratio = sparsity_ratio
Z
zhangjinchao01 已提交
3342
        return hook
3343 3344 3345 3346
    elif type == 'dpruning':
        hook = ParameterUpdaterHookConfig()
        hook.type = type
        return hook
Z
zhangjinchao01 已提交
3347 3348 3349 3350 3351
    else:
        return None


@config_func
Q
qijun 已提交
3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372
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
3373 3374
              update_hooks=None,
              initializer=None):
Z
zhangjinchao01 已提交
3375 3376 3377 3378 3379 3380 3381

    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
3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392
    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
3393 3394
    config_assert(not momentum or not decay_rate_l1,
                  "momentum and decay_rate_l1 cannot both be non-zero")
3395 3396 3397 3398 3399

    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
3400 3401 3402 3403
    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)
3404

Q
qijun 已提交
3405 3406
    num_batches_regularization = default(num_batches_regularization,
                                         g_default_num_batches_regularization)
3407 3408 3409
    if num_batches_regularization is not None:
        para.num_batches_regularization = int(num_batches_regularization)

Z
zhangjinchao01 已提交
3410 3411 3412 3413 3414 3415
    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
3416 3417
    gradient_clipping_threshold = default(gradient_clipping_threshold,
                                          g_default_gradient_clipping_threshold)
3418 3419
    if gradient_clipping_threshold is not None:
        para.gradient_clipping_threshold = gradient_clipping_threshold
Q
qijun 已提交
3420 3421
    para.initial_strategy = default(initial_strategy,
                                    g_default_initial_strategy)
Z
zhangjinchao01 已提交
3422 3423 3424 3425 3426 3427
    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
3428 3429 3430
            print(
                "Use initial_smart, but dims not set. Initial_smart may not be used in this layer"
            )
Z
zhangjinchao01 已提交
3431 3432 3433 3434
            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)
3435 3436 3437 3438 3439 3440 3441

    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
3442 3443 3444 3445
    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")
3446 3447
    if is_shared is not None:
        para.is_shared = is_shared
Z
zhangjinchao01 已提交
3448 3449 3450 3451 3452

    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
3453
            update_hooks = update_hooks()
Z
zhangjinchao01 已提交
3454 3455 3456 3457 3458

        if isinstance(update_hooks, list):
            for hook in update_hooks:
                para.update_hooks.extend([hook])
        else:
X
xzl 已提交
3459
            para.update_hooks.extend([update_hooks])
Z
zhangjinchao01 已提交
3460 3461

    g_parameter_map[name] = para
X
xuwei06 已提交
3462 3463 3464 3465 3466
    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
3467 3468 3469 3470 3471 3472 3473


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

Q
qijun 已提交
3474

Z
zhangjinchao01 已提交
3475 3476 3477 3478 3479
@config_func
def default_initial_mean(val):
    global g_default_initial_mean
    g_default_initial_mean = val

Q
qijun 已提交
3480

Z
zhangjinchao01 已提交
3481 3482 3483 3484 3485
@config_func
def default_initial_strategy(val):
    global g_default_initial_strategy
    g_default_initial_strategy = val

Q
qijun 已提交
3486

Z
zhangjinchao01 已提交
3487 3488 3489 3490 3491
@config_func
def default_initial_smart(val):
    global g_default_initial_smart
    g_default_initial_smart = val

Q
qijun 已提交
3492

Z
zhangjinchao01 已提交
3493 3494 3495 3496 3497
@config_func
def default_momentum(val):
    global g_default_momentum
    g_default_momentum = val

Q
qijun 已提交
3498

Z
zhangjinchao01 已提交
3499 3500 3501 3502 3503
@config_func
def default_decay_rate(val):
    global g_default_decay_rate
    g_default_decay_rate = val

Q
qijun 已提交
3504

Z
zhangjinchao01 已提交
3505 3506 3507 3508 3509
@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
3510

Z
zhangjinchao01 已提交
3511 3512 3513 3514 3515
@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
3516

Z
zhangjinchao01 已提交
3517 3518 3519 3520 3521
@config_func
def default_device(val):
    global g_default_device
    g_default_device = val

Q
qijun 已提交
3522

Z
zhangjinchao01 已提交
3523 3524 3525 3526 3527
@config_func
def default_update_hooks(val):
    global g_default_update_hooks
    g_default_update_hooks = val

Q
qijun 已提交
3528

Z
zhangjinchao01 已提交
3529 3530 3531 3532 3533
@config_func
def default_compact_func(val):
    global g_default_compact_func
    g_default_compact_func = val

Q
qijun 已提交
3534

Z
zhangjinchao01 已提交
3535 3536 3537 3538 3539
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
3540 3541 3542
        execfile(config_file,
                 make_config_environment(config_file, config_args), local_args)

Z
zhangjinchao01 已提交
3543 3544
    return Import

Q
qijun 已提交
3545

X
xuwei06 已提交
3546
DEFAULT_SETTING = dict(
Z
zhangjinchao01 已提交
3547 3548 3549 3550 3551
    batch_size=None,
    mini_batch_size=None,
    algorithm='async_sgd',
    async_lagged_grad_discard_ratio=1.5,
    learning_method='momentum',
3552
    gradient_clipping_threshold=None,
Z
zhangjinchao01 已提交
3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574
    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
3575 3576 3577
    adam_beta1=0.9,
    adam_beta2=0.999,
    adam_epsilon=1e-8, )
Z
zhangjinchao01 已提交
3578

X
xuwei06 已提交
3579
settings = copy.deepcopy(DEFAULT_SETTING)
X
xuwei06 已提交
3580

Q
qijun 已提交
3581
settings_deprecated = dict(usage_ratio=1., )
Z
zhangjinchao01 已提交
3582 3583 3584 3585

trainer_settings = dict(
    save_dir="./output/model",
    init_model_path=None,
Q
qijun 已提交
3586 3587
    start_pass=0, )

Z
zhangjinchao01 已提交
3588 3589 3590 3591 3592

@config_func
def Settings(**args):
    for k, v in args.iteritems():
        if k == "usage_ratio":
Q
qijun 已提交
3593 3594
            logger.warning(
                "Deprecated: define usage_ratio in DataConfig instead")
Z
zhangjinchao01 已提交
3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605
            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
3606

Z
zhangjinchao01 已提交
3607 3608 3609 3610
@config_func
def cluster_config(**args):
    pass

Q
qijun 已提交
3611

Z
zhangjinchao01 已提交
3612 3613 3614 3615 3616 3617 3618 3619 3620
@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
3621

Z
zhangjinchao01 已提交
3622 3623 3624 3625
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
3626

Z
zhangjinchao01 已提交
3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641
        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
3642
        get_config_arg=make_get_config_arg(config_args), )
Z
zhangjinchao01 已提交
3643 3644 3645 3646 3647

    funcs.update(g_extended_config_funcs)

    return funcs

Q
qijun 已提交
3648

Z
zhangjinchao01 已提交
3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664
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
3665

Z
zhangjinchao01 已提交
3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677
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
3678

Z
zhangjinchao01 已提交
3679 3680 3681 3682
def my_fatal(s):
    logger.critical(s)
    raise Exception()

Y
Yu Yang 已提交
3683

3684
_parse_config_hooks = set()
Y
Yu Yang 已提交
3685 3686


3687 3688 3689 3690 3691 3692 3693
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
3694

Y
Yu Yang 已提交
3695

3696
def update_g_config():
Z
zhangjinchao01 已提交
3697
    '''
3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720
    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


3721
def begin_parse():
Z
zhangjinchao01 已提交
3722
    init_config_environment()
3723 3724
    for hook in _parse_config_hooks:
        hook()
Z
zhangjinchao01 已提交
3725 3726 3727 3728 3729

    logger.findCaller = find_caller
    logger.fatal = my_fatal

    g_config.model_config.type = "nn"
X
xuwei06 已提交
3730 3731 3732 3733 3734 3735 3736 3737 3738

    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):
3739 3740 3741 3742
    '''
    @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
3743

3744
    begin_parse()
X
xuwei06 已提交
3745 3746
    config_args = {}

Z
zhangjinchao01 已提交
3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758
    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)

3759 3760
    if hasattr(trainer_config, '__call__'):
        trainer_config.func_globals.update(
L
Luo Tao 已提交
3761
            make_config_environment("", config_args))
3762
        trainer_config()
H
hanchao 已提交
3763
    else:
3764 3765
        execfile(trainer_config,
                 make_config_environment(trainer_config, config_args))
Z
zhangjinchao01 已提交
3766

3767
    return update_g_config()
Z
zhangjinchao01 已提交
3768 3769


3770
def parse_config_and_serialize(trainer_config, config_arg_str):
Z
zhangjinchao01 已提交
3771
    try:
3772
        config = parse_config(trainer_config, config_arg_str)
Z
zhangjinchao01 已提交
3773 3774 3775 3776 3777 3778
        #logger.info(config)
        return config.SerializeToString()
    except:
        traceback.print_exc()
        raise

Q
qijun 已提交
3779

Z
zhangjinchao01 已提交
3780 3781 3782 3783 3784 3785 3786 3787
if __name__ == '__main__':
    try:
        config = parse_config(sys.argv[1], '')
        config.SerializeToString()
        __real_print__(str(config))
    except:
        traceback.print_exc()
        raise