import torch
from torch import nn
from d2l import torch as d2l


# NiN块
def nin_block(in_channels, out_channels, kernel_size, strides, padding):
    return nn.Sequential(
        nn.Conv2d(in_channels, out_channels, kernel_size, strides, padding),
        nn.ReLU(),
        nn.Conv2d(out_channels, out_channels, kernel_size=1, stride=1),
        nn.ReLU(),
        nn.Conv2d(out_channels, out_channels, kernel_size=1, stride=1),
        nn.ReLU()
    )


# NiN模型
net = nn.Sequential(
    nin_block(1, 96, kernel_size=11, strides=4, padding=0),
    nn.MaxPool2d(3, 2),
    nin_block(96, 256, kernel_size=5, strides=1, padding=2),
    nn.MaxPool2d(3, 2),
    nin_block(256, 384, kernel_size=3, strides=1, padding=1),
    nn.MaxPool2d(3, 2), nn.Dropout(0.5),
    # 标签类别数是10
    nin_block(384, 10, kernel_size=3, strides=1, padding=1),
    nn.AdaptiveAvgPool2d((1, 1)),
    # 将四维的输出转成二维的输出
    nn.Flatten()
)

# 训练模型
lr, num_epochs, batch_size = 0.03, 50, 128
train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size, resize=224)
d2l.train_ch6(net, train_iter, test_iter, num_epochs, lr, d2l.try_gpu())
d2l.plt.show()

Output

loss 0.662, train acc 0.800, test acc 0.798
2038.3 examples/sec on cuda:0

有一说一，NiN跑了50轮的表现还不如AlexNet十轮的表现…可能数据集选的不对吧