【pytorch】手写softmax回归单层网络模型范例 + 利用pytorch模块简单实现softmax回归

一、手写softmax回归单层网络模型

（1）代码

import sys

import matplotlib.pyplot as plt
import numpy as np
import torch
import torchvision
import torchvision.transforms as transforms
from IPython import display


def use_svg_display():
    """Use svg format to display plot in jupyter"""
    display.set_matplotlib_formats('svg')


def sgd(params, lr, batch_size):
    # 为了和原书保持一致，这里除以了batch_size，但是应该是不用除的，因为一般用PyTorch计算loss时就默认已经
    # 沿batch维求了平均了。
    for param in params:
        param.data -= lr * param.grad / batch_size  # 注意这里更改param时用的param.data


def show_fashion_mnist(images, labels):
    use_svg_display()
    # 这里的_表示我们忽略（不使用）的变量
    _, figs = plt.subplots(1, len(images), figsize=(12, 12))
    for f, img, lbl in zip(figs, images, labels):
        f.imshow(img.view((28, 28)).numpy())
        f.set_title(lbl)
        f.axes.get_xaxis().set_visible(False)
        f.axes.get_yaxis().set_visible(False)
    plt.show()


def load_data_fashion_mnist(batch_size, root='./data'):
    """Download the fashion mnist dataset and then load into memory."""
    transform = transforms.ToTensor()
    mnist_train = torchvision.datasets.FashionMNIST(root=root, train=True, download=True, transform=transform)
    mnist_test = torchvision.datasets.FashionMNIST(root=root, train=False, download=True, transform=transform)
    if sys.platform.startswith('win'):
        num_workers = 0  # 0表示不用额外的进程来加速读取数据
    else:
        num_workers = 4
    train_iter = torch.utils.data.DataLoader(mnist_train, batch_size=batch_size, shuffle=True, num_workers=num_workers)
    test_iter = torch.utils.data.DataLoader(mnist_test, batch_size=batch_size, shuffle=False, num_workers=num_workers)

    return train_iter, test_iter


def get_fashion_mnist_labels(labels):
    text_labels = ['t-shirt', 'trouser', 'pullover', 'dress', 'coat',
                   'sandal', 'shirt', 'sneaker', 'bag', 'ankle boot']
    return [text_labels[int(i)] for i in labels]


def softmax(X):
    X_exp = X.exp()
    # 同一列（dim=0）或同一行（dim=1）
    partition = X_exp.sum(dim=1, keepdim=True)
    return X_exp / partition  # 这里应用了广播机制


def cross_entropy(y_hat, y):
    return - torch.log(y_hat.gather(dim=1, index=y.view(-1, 1)))
    # return - torch.log(torch.gather(y_hat, dim=1, index=y))


def accuracy(y_hat, y):
    return (y_hat.argmax(dim=1) == y).float().mean().item()


def net(X):
    return softmax(torch.mm(X.view((-1, num_inputs)), W) + b)


def evaluate_accuracy(data_iter):
    acc_sum, n = 0.0, 0
    for X, y in data_iter:
        # print(X.shape)
        # print(y.shape)
        acc_sum += (net(X).argmax(dim=1) == y).float().sum().item()
        n += y.shape[0]
    return acc_sum / n


# 本函数已保存在d2lzh包中方便以后使用
def train_ch3(net, train_iter, test_iter, loss, num_epochs, batch_size,
              params=None, lr=None, optimizer=None):
    for epoch in range(num_epochs):
        train_l_sum, train_acc_sum, n = 0.0, 0.0, 0
        for X, y in train_iter:
            y_hat = net(X)
            l = loss(y_hat, y).sum()
            # 梯度清零
            if optimizer is not None:
                optimizer.zero_grad()
            elif params is not None and params[0].grad is not None:
                for param in params:
                    param.grad.data.zero_()

            l.backward()
            if optimizer is None:
                sgd(params, lr, batch_size)
            else:
                optimizer.step()  # “softmax回归的简洁实现”一节将用到

            train_l_sum += l.item()
            train_acc_sum += (y_hat.argmax(dim=1) == y).sum().item()
            n += y.shape[0]
        test_acc = evaluate_accuracy(test_iter)
        print('epoch %d, loss %.4f, train acc %.3f, test acc %.3f'
              % (epoch + 1, train_l_sum / n, train_acc_sum / n, test_acc))


if __name__ == '__main__':
    batch_size = 256
    train_iter, test_iter = load_data_fashion_mnist(batch_size)

    num_inputs = 784
    num_outputs = 10

    W = torch.tensor(np.random.normal(0, 0.01, (num_inputs, num_outputs)), dtype=torch.float)
    b = torch.zeros(num_outputs, dtype=torch.float)
    W.requires_grad_(requires_grad=True)
    b.requires_grad_(requires_grad=True)

    num_epochs, lr = 5, 0.1
    train_ch3(net, train_iter, test_iter, cross_entropy, num_epochs, batch_size, [W, b], lr)

    # 预测
    X, y = iter(test_iter).next()

    true_labels = get_fashion_mnist_labels(y.numpy())
    pred_labels = get_fashion_mnist_labels(net(X).argmax(dim=1).numpy())
    titles = [true + '\n' + pred for true, pred in zip(true_labels, pred_labels)]

    show_fashion_mnist(X[0:9], titles[0:9])

（2）结果

epoch 1, loss 0.7861, train acc 0.748, test acc 0.793
epoch 2, loss 0.5697, train acc 0.813, test acc 0.811
epoch 3, loss 0.5247, train acc 0.826, test acc 0.822
epoch 4, loss 0.5015, train acc 0.833, test acc 0.826
epoch 5, loss 0.4856, train acc 0.837, test acc 0.825

二、利用pytorch模块简单实现softmax回归

（1）代码

import sys
from collections import OrderedDict

import torch
import torchvision
from torch import nn
from torch.nn import init


class FlattenLayer(nn.Module):
    def __init__(self):
        super(FlattenLayer, self).__init__()

    def forward(self, x):  # x shape: (batch, *, *, ...)
        return x.view(x.shape[0], -1)


def sgd(params, lr, batch_size):
    # 为了和原书保持一致，这里除以了batch_size，但是应该是不用除的，因为一般用PyTorch计算loss时就默认已经
    # 沿batch维求了平均了。
    for param in params:
        param.data -= lr * param.grad / batch_size  # 注意这里更改param时用的param.data


def evaluate_accuracy(data_iter, net, device=None):
    if device is None and isinstance(net, torch.nn.Module):
        # 如果没指定device就使用net的device
        device = list(net.parameters())[0].device
    acc_sum, n = 0.0, 0
    with torch.no_grad():
        for X, y in data_iter:
            if isinstance(net, torch.nn.Module):
                net.eval()  # 评估模式, 这会关闭dropout
                acc_sum += (net(X.to(device)).argmax(dim=1) == y.to(device)).float().sum().cpu().item()
                net.train()  # 改回训练模式
            else:  # 自定义的模型, 3.13节之后不会用到, 不考虑GPU
                if ('is_training' in net.__code__.co_varnames):  # 如果有is_training这个参数
                    # 将is_training设置成False
                    acc_sum += (net(X, is_training=False).argmax(dim=1) == y).float().sum().item()
                else:
                    acc_sum += (net(X).argmax(dim=1) == y).float().sum().item()
            n += y.shape[0]
    return acc_sum / n


def train_ch3(net, train_iter, test_iter, loss, num_epochs, batch_size,
              params=None, lr=None, optimizer=None):
    for epoch in range(num_epochs):
        train_l_sum, train_acc_sum, n = 0.0, 0.0, 0
        for X, y in train_iter:
            y_hat = net(X)
            l = loss(y_hat, y).sum()

            # 梯度清零
            if optimizer is not None:
                optimizer.zero_grad()
            elif params is not None and params[0].grad is not None:
                for param in params:
                    param.grad.data.zero_()

            l.backward()
            if optimizer is None:
                sgd(params, lr, batch_size)
            else:
                optimizer.step()  # “softmax回归的简洁实现”一节将用到

            train_l_sum += l.item()
            train_acc_sum += (y_hat.argmax(dim=1) == y).sum().item()
            n += y.shape[0]
        test_acc = evaluate_accuracy(test_iter, net)
        print('epoch %d, loss %.4f, train acc %.3f, test acc %.3f'
              % (epoch + 1, train_l_sum / n, train_acc_sum / n, test_acc))


def load_data_fashion_mnist(batch_size, resize=None, root='./data'):
    """Download the fashion mnist dataset and then load into memory."""
    trans = []
    if resize:
        trans.append(torchvision.transforms.Resize(size=resize))
    trans.append(torchvision.transforms.ToTensor())

    transform = torchvision.transforms.Compose(trans)
    mnist_train = torchvision.datasets.FashionMNIST(root=root, train=True, download=True, transform=transform)
    mnist_test = torchvision.datasets.FashionMNIST(root=root, train=False, download=True, transform=transform)
    if sys.platform.startswith('win'):
        num_workers = 0  # 0表示不用额外的进程来加速读取数据
    else:
        num_workers = 4
    train_iter = torch.utils.data.DataLoader(mnist_train, batch_size=batch_size, shuffle=True, num_workers=num_workers)
    test_iter = torch.utils.data.DataLoader(mnist_test, batch_size=batch_size, shuffle=False, num_workers=num_workers)

    return train_iter, test_iter


if __name__ == '__main__':
    num_inputs = 784
    num_outputs = 10

    net = nn.Sequential(
        # FlattenLayer(),
        # nn.Linear(num_inputs, num_outputs)
        OrderedDict([
            ('flatten', FlattenLayer()),
            ('linear', nn.Linear(num_inputs, num_outputs))])
    )

    batch_size = 256
    train_iter, test_iter = load_data_fashion_mnist(batch_size)

    init.normal_(net.linear.weight, mean=0, std=0.01)
    init.constant_(net.linear.bias, val=0)

    loss = nn.CrossEntropyLoss()

    # print([i for i in net.parameters()])

    optimizer = torch.optim.SGD(net.parameters(), lr=0.1)

    num_epochs = 5
    train_ch3(net, train_iter, test_iter, loss, num_epochs, batch_size, None, None, optimizer)

（2）结果

epoch 1, loss 0.0031, train acc 0.750, test acc 0.784
epoch 2, loss 0.0022, train acc 0.813, test acc 0.811
epoch 3, loss 0.0021, train acc 0.825, test acc 0.817
epoch 4, loss 0.0020, train acc 0.833, test acc 0.819
epoch 5, loss 0.0019, train acc 0.837, test acc 0.825