Networks Using Blocks (VGG)¶

In [1]:

import d2l
from mxnet import gluon, np, npx
from mxnet.gluon import nn
npx.set_np()

train_iter, test_iter = d2l.load_data_fashion_mnist(
    batch_size=128, resize=224)

VGG block

In [2]:

def vgg_block(num_convs, num_channels):
    blk = nn.Sequential()
    for _ in range(num_convs):
        blk.add(nn.Conv2D(num_channels, kernel_size=3,
                          padding=1, activation='relu'))
    blk.add(nn.MaxPool2D(pool_size=2, strides=2))
    return blk

The model

In [3]:

conv_arch = ((1, 64), (1, 128), (2, 256), (2, 512), (2, 512))

def vgg(conv_arch):
    net = nn.Sequential()
    # The convolutional layer part
    for (num_convs, num_channels) in conv_arch:
        net.add(vgg_block(num_convs, num_channels))
    # The fully connected layer part
    net.add(nn.Dense(4096, activation='relu'), nn.Dropout(0.5),
            nn.Dense(4096, activation='relu'), nn.Dropout(0.5),
            nn.Dense(10))
    return net

net = vgg(conv_arch)