test_imperative_se_resnext.py 16.4 KB
Newer Older
Y
Yan Xu 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58
# Copyright (c) 2018 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import contextlib
import unittest
import numpy as np
import six

import paddle
import paddle.fluid as fluid
from paddle.fluid import core
from paddle.fluid.layer_helper import LayerHelper
from paddle.fluid.dygraph.nn import Conv2D, Pool2D, BatchNorm, FC
from paddle.fluid.dygraph.base import to_variable
from test_imperative_base import new_program_scope

batch_size = 8
train_parameters = {
    "input_size": [3, 224, 224],
    "input_mean": [0.485, 0.456, 0.406],
    "input_std": [0.229, 0.224, 0.225],
    "learning_strategy": {
        "name": "piecewise_decay",
        "batch_size": batch_size,
        "epochs": [30, 60, 90],
        "steps": [0.1, 0.01, 0.001, 0.0001]
    },
    "batch_size": batch_size,
    "lr": 0.1,
    "total_images": 6149,
}


def optimizer_setting(params):
    ls = params["learning_strategy"]
    if ls["name"] == "piecewise_decay":
        if "total_images" not in params:
            total_images = 6149
        else:
            total_images = params["total_images"]
        # TODO(Yancey1989): using lr decay if it is ready.
        #batch_size = ls["batch_size"]
        #step = int(total_images / batch_size + 1)

        #bd = [step * e for e in ls["epochs"]]
        #base_lr = params["lr"]
        #lr = [base_lr * (0.1**i) for i in range(len(bd) + 1)]
Y
Yan Xu 已提交
59
        optimizer = fluid.optimizer.SGD(learning_rate=0.01)
Y
Yan Xu 已提交
60 61 62 63 64 65

    return optimizer


class ConvBNLayer(fluid.dygraph.Layer):
    def __init__(self,
66
                 num_channels,
Y
Yan Xu 已提交
67 68 69 70 71
                 num_filters,
                 filter_size,
                 stride=1,
                 groups=1,
                 act=None):
72
        super(ConvBNLayer, self).__init__()
Y
Yan Xu 已提交
73 74

        self._conv = Conv2D(
75
            num_channels=num_channels,
Y
Yan Xu 已提交
76 77 78 79 80 81 82 83
            num_filters=num_filters,
            filter_size=filter_size,
            stride=stride,
            padding=(filter_size - 1) // 2,
            groups=groups,
            act=None,
            bias_attr=None)

84
        self._batch_norm = BatchNorm(num_filters, act=act)
Y
Yan Xu 已提交
85 86 87 88 89 90 91 92 93 94 95 96

    def forward(self, inputs):
        y = self._conv(inputs)
        y = self._batch_norm(y)

        return y


class SqueezeExcitation(fluid.dygraph.Layer):
    def __init__(self, name_scope, num_channels, reduction_ratio):

        super(SqueezeExcitation, self).__init__(name_scope)
97
        self._pool = Pool2D(pool_size=0, pool_type='avg', global_pooling=True)
Y
Yan Xu 已提交
98 99 100 101 102 103 104 105 106 107 108
        self._squeeze = FC(
            self.full_name(),
            size=num_channels // reduction_ratio,
            param_attr=fluid.ParamAttr(
                initializer=fluid.initializer.Constant(value=0.05)),
            act='relu')
        self._excitation = FC(
            self.full_name(),
            size=num_channels,
            param_attr=fluid.ParamAttr(
                initializer=fluid.initializer.Constant(value=0.05)),
Y
Yan Xu 已提交
109
            act='sigmoid')
Y
Yan Xu 已提交
110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126

    def forward(self, input):
        y = self._pool(input)
        y = self._squeeze(y)
        y = self._excitation(y)
        y = fluid.layers.elementwise_mul(x=input, y=y, axis=0)
        return y


class BottleneckBlock(fluid.dygraph.Layer):
    def __init__(self,
                 num_channels,
                 num_filters,
                 stride,
                 cardinality,
                 reduction_ratio,
                 shortcut=True):
127
        super(BottleneckBlock, self).__init__()
Y
Yan Xu 已提交
128 129

        self.conv0 = ConvBNLayer(
130
            num_channels=num_channels, num_filters=num_filters, filter_size=1)
Y
Yan Xu 已提交
131
        self.conv1 = ConvBNLayer(
132
            num_channels=num_filters,
Y
Yan Xu 已提交
133 134 135 136 137
            num_filters=num_filters,
            filter_size=3,
            stride=stride,
            groups=cardinality)
        self.conv2 = ConvBNLayer(
138
            num_channels=num_filters,
Y
Yan Xu 已提交
139 140 141 142 143 144 145 146 147 148 149
            num_filters=num_filters * 4,
            filter_size=1,
            act='relu')

        self.scale = SqueezeExcitation(
            self.full_name(),
            num_channels=num_filters * 4,
            reduction_ratio=reduction_ratio)

        if not shortcut:
            self.short = ConvBNLayer(
150
                num_channels=num_channels,
Y
Yan Xu 已提交
151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191
                num_filters=num_filters * 4,
                filter_size=1,
                stride=stride)

        self.shortcut = shortcut

        self._num_channels_out = num_filters * 4

    def forward(self, inputs):
        y = self.conv0(inputs)
        conv1 = self.conv1(y)
        conv2 = self.conv2(conv1)
        scale = self.scale(conv2)

        if self.shortcut:
            short = inputs
        else:
            short = self.short(inputs)

        y = fluid.layers.elementwise_add(x=short, y=scale)

        layer_helper = LayerHelper(self.full_name(), act='relu')
        y = layer_helper.append_activation(y)
        return y


class SeResNeXt(fluid.dygraph.Layer):
    def __init__(self, name_scope, layers=50, class_dim=102):
        super(SeResNeXt, self).__init__(name_scope)

        self.layers = layers
        supported_layers = [50, 101, 152]
        assert layers in supported_layers, \
            "supported layers are {} but input layer is {}".format(supported_layers, layers)

        if layers == 50:
            cardinality = 32
            reduction_ratio = 16
            depth = [3, 4, 6, 3]
            num_filters = [128, 256, 512, 1024]
            self.conv0 = ConvBNLayer(
192
                num_channels=3,
Y
Yan Xu 已提交
193 194 195 196 197
                num_filters=64,
                filter_size=7,
                stride=2,
                act='relu')
            self.pool = Pool2D(
198
                pool_size=3, pool_stride=2, pool_padding=1, pool_type='max')
Y
Yan Xu 已提交
199 200 201 202 203 204
        elif layers == 101:
            cardinality = 32
            reduction_ratio = 16
            depth = [3, 4, 23, 3]
            num_filters = [128, 256, 512, 1024]
            self.conv0 = ConvBNLayer(
205
                num_channels=3,
Y
Yan Xu 已提交
206 207 208 209 210
                num_filters=3,
                filter_size=7,
                stride=2,
                act='relu')
            self.pool = Pool2D(
211
                pool_size=3, pool_stride=2, pool_padding=1, pool_type='max')
Y
Yan Xu 已提交
212 213 214 215 216 217
        elif layers == 152:
            cardinality = 64
            reduction_ratio = 16
            depth = [3, 8, 36, 3]
            num_filters = [128, 256, 512, 1024]
            self.conv0 = ConvBNLayer(
218
                num_channels=3,
Y
Yan Xu 已提交
219 220 221 222 223
                num_filters=3,
                filter_size=7,
                stride=2,
                act='relu')
            self.conv1 = ConvBNLayer(
224
                num_channels=3,
Y
Yan Xu 已提交
225 226 227 228 229
                num_filters=3,
                filter_size=7,
                stride=2,
                act='relu')
            self.conv2 = ConvBNLayer(
230
                num_channels=7,
Y
Yan Xu 已提交
231 232 233 234 235
                num_filters=3,
                filter_size=7,
                stride=2,
                act='relu')
            self.pool = Pool2D(
236
                pool_size=3, pool_stride=2, pool_padding=1, pool_type='max')
Y
Yan Xu 已提交
237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256

        self.bottleneck_block_list = []
        num_channels = 64
        for block in range(len(depth)):
            shortcut = False
            for i in range(depth[block]):
                bottleneck_block = self.add_sublayer(
                    'bb_%d_%d' % (block, i),
                    BottleneckBlock(
                        num_channels=num_channels,
                        num_filters=num_filters[block],
                        stride=2 if i == 0 and block != 0 else 1,
                        cardinality=cardinality,
                        reduction_ratio=reduction_ratio,
                        shortcut=shortcut))
                num_channels = bottleneck_block._num_channels_out
                self.bottleneck_block_list.append(bottleneck_block)
                shortcut = True

        self.pool2d_avg = Pool2D(
257
            pool_size=7, pool_type='avg', global_pooling=True)
Y
Yan Xu 已提交
258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285
        import math
        stdv = 1.0 / math.sqrt(2048 * 1.0)

        self.out = FC(self.full_name(),
                      size=class_dim,
                      act='softmax',
                      param_attr=fluid.param_attr.ParamAttr(
                          initializer=fluid.initializer.Uniform(-stdv, stdv)))

    def forward(self, inputs):
        if self.layers == 50 or self.layers == 101:
            y = self.conv0(inputs)
            y = self.pool(y)
        elif self.layers == 152:
            y = self.conv0(inputs)
            y = self.conv1(inputs)
            y = self.conv2(inputs)
            y = self.pool(y)

        for bottleneck_block in self.bottleneck_block_list:
            y = bottleneck_block(y)
        y = self.pool2d_avg(y)
        y = fluid.layers.dropout(y, dropout_prob=0.2)
        y = self.out(y)
        return y


class TestImperativeResneXt(unittest.TestCase):
286 287 288 289 290 291 292 293 294
    def reader_decorator(self, reader):
        def _reader_imple():
            for item in reader():
                doc = np.array(item[0]).reshape(3, 224, 224)
                label = np.array(item[1]).astype('int64').reshape(1)
                yield doc, label

        return _reader_imple

Y
Yan Xu 已提交
295 296 297 298
    def test_se_resnext_float32(self):
        seed = 90

        batch_size = train_parameters["batch_size"]
299
        batch_num = 1
Y
Yan Xu 已提交
300
        epoch_num = 1
Y
Yan Xu 已提交
301 302 303 304 305 306 307 308 309
        with fluid.dygraph.guard():
            fluid.default_startup_program().random_seed = seed
            fluid.default_main_program().random_seed = seed

            se_resnext = SeResNeXt("se_resnext")
            optimizer = optimizer_setting(train_parameters)
            np.random.seed(seed)
            import random
            random.seed = seed
310 311 312 313 314 315 316 317 318

            batch_py_reader = fluid.io.PyReader(capacity=1)
            batch_py_reader.decorate_sample_list_generator(
                paddle.batch(
                    self.reader_decorator(
                        paddle.dataset.flowers.train(use_xmap=False)),
                    batch_size=batch_size,
                    drop_last=True),
                places=fluid.CPUPlace())
Y
Yan Xu 已提交
319 320 321

            dy_param_init_value = {}
            for param in se_resnext.parameters():
L
lujun 已提交
322
                dy_param_init_value[param.name] = param.numpy()
Y
Yan Xu 已提交
323
            for epoch_id in range(epoch_num):
324
                for batch_id, data in enumerate(batch_py_reader()):
Y
Yan Xu 已提交
325 326 327 328

                    if batch_id >= batch_num and batch_num != -1:
                        break

329 330 331
                    img = data[0]
                    label = data[1]
                    label.stop_gradient = True
L
lujun 已提交
332
                    label.stop_gradient = True
Y
Yan Xu 已提交
333 334 335 336 337

                    out = se_resnext(img)
                    loss = fluid.layers.cross_entropy(input=out, label=label)
                    avg_loss = fluid.layers.mean(x=loss)

L
lujun 已提交
338
                    dy_out = avg_loss.numpy()
Y
Yan Xu 已提交
339 340 341 342

                    if batch_id == 0:
                        for param in se_resnext.parameters():
                            if param.name not in dy_param_init_value:
L
lujun 已提交
343 344
                                dy_param_init_value[param.name] = param.numpy()
                    avg_loss.backward()
Y
Yan Xu 已提交
345 346 347 348

                    #dy_grad_value = {}
                    #for param in se_resnext.parameters():
                    #    if param.trainable:
349
                    #        np_array = np.array(param._grad_ivar().value()
Y
Yan Xu 已提交
350 351 352 353 354 355 356
                    #                            .get_tensor())
                    #        dy_grad_value[param.name + core.grad_var_suffix()] = np_array

                    optimizer.minimize(avg_loss)
                    se_resnext.clear_gradients()

                    dy_param_value = {}
Y
Yan Xu 已提交
357
                    for param in se_resnext.parameters():
L
lujun 已提交
358
                        dy_param_value[param.name] = param.numpy()
Y
Yan Xu 已提交
359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374

        with new_program_scope():
            fluid.default_startup_program().random_seed = seed
            fluid.default_main_program().random_seed = seed

            exe = fluid.Executor(fluid.CPUPlace(
            ) if not core.is_compiled_with_cuda() else fluid.CUDAPlace(0))

            se_resnext = SeResNeXt("se_resnext")
            optimizer = optimizer_setting(train_parameters)

            np.random.seed(seed)
            import random
            random.seed = seed
            train_reader = paddle.batch(
                paddle.dataset.flowers.train(use_xmap=False),
Y
Yan Xu 已提交
375 376
                batch_size=batch_size,
                drop_last=True)
Y
Yan Xu 已提交
377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401

            img = fluid.layers.data(
                name='pixel', shape=[3, 224, 224], dtype='float32')
            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
            out = se_resnext(img)
            loss = fluid.layers.cross_entropy(input=out, label=label)
            avg_loss = fluid.layers.mean(x=loss)
            optimizer.minimize(avg_loss)

            # initialize params and fetch them
            static_param_init_value = {}
            static_param_name_list = []
            static_grad_name_list = []
            for param in se_resnext.parameters():
                static_param_name_list.append(param.name)
            for param in se_resnext.parameters():
                if param.trainable:
                    static_grad_name_list.append(param.name +
                                                 core.grad_var_suffix())

            out = exe.run(fluid.default_startup_program(),
                          fetch_list=static_param_name_list)

            for i in range(len(static_param_name_list)):
                static_param_init_value[static_param_name_list[i]] = out[i]
Y
Yan Xu 已提交
402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437
            for epoch_id in range(epoch_num):
                for batch_id, data in enumerate(train_reader()):
                    if batch_id >= batch_num and batch_num != -1:
                        break

                    static_x_data = np.array(
                        [x[0].reshape(3, 224, 224)
                         for x in data]).astype('float32')
                    y_data = np.array(
                        [x[1] for x in data]).astype('int64').reshape(
                            [batch_size, 1])

                    fetch_list = [avg_loss.name]
                    fetch_list.extend(static_param_name_list)
                    fetch_list.extend(static_grad_name_list)
                    out = exe.run(
                        fluid.default_main_program(),
                        feed={"pixel": static_x_data,
                              "label": y_data},
                        fetch_list=fetch_list)

                    static_param_value = {}
                    static_grad_value = {}
                    static_out = out[0]
                    param_start_pos = 1
                    grad_start_pos = len(
                        static_param_name_list) + param_start_pos
                    for i in range(
                            param_start_pos,
                            len(static_param_name_list) + param_start_pos):
                        static_param_value[static_param_name_list[
                            i - param_start_pos]] = out[i]
                    for i in range(grad_start_pos,
                                   len(static_grad_name_list) + grad_start_pos):
                        static_grad_value[static_grad_name_list[
                            i - grad_start_pos]] = out[i]
Y
Yan Xu 已提交
438 439 440 441 442 443 444 445
        self.assertTrue(np.allclose(static_out, dy_out))

        self.assertEqual(len(dy_param_init_value), len(static_param_init_value))

        for key, value in six.iteritems(static_param_init_value):
            self.assertTrue(np.allclose(value, dy_param_init_value[key]))
            self.assertTrue(np.isfinite(value.all()))
            self.assertFalse(np.isnan(value.any()))
Y
Yan Xu 已提交
446 447 448 449 450 451
        # FIXME(Yancey1989): np.array(_ivar.value().get_tensor()) leads to memory lake
        #self.assertEqual(len(dy_grad_value), len(static_grad_value))
        #for key, value in six.iteritems(static_grad_value):
        #    self.assertTrue(np.allclose(value, dy_grad_value[key]))
        #    self.assertTrue(np.isfinite(value.all()))
        #    self.assertFalse(np.isnan(value.any()))
Y
Yan Xu 已提交
452 453 454 455 456 457 458 459 460 461

        self.assertEqual(len(dy_param_value), len(static_param_value))
        for key, value in six.iteritems(static_param_value):
            self.assertTrue(np.allclose(value, dy_param_value[key]))
            self.assertTrue(np.isfinite(value.all()))
            self.assertFalse(np.isnan(value.any()))


if __name__ == '__main__':
    unittest.main()