多层感知机的实现

理顺几个关键组件：

• 我需要训练模型，首先就需要构建模型神经网络net,需要params（W，b）、relu层.
• 使用什么训练。需要加载数据datasets
• 怎么训练。需要损失函数计算预测与实际的差距loss,知道差距之后，怎么减少损失靠近实际，需要优化函数来更新参数optim.
• 训练参数。一次训练多少样本batch_size,需要训练多少轮num_epochs,学习率lr

手搓版本

datasets

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

batch_size = 256
train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size)

parameters

num_inputs, num_outputs, num_hidden = 784, 10, 256

W1 = nn.Parameter(torch.randn(
    num_inputs, num_hidden, requires_grad=True) * 0.01)
b1 = nn.Parameter(torch.zeros(num_hidden, requires_grad=True))
W2 = nn.Parameter(torch.randn(
    num_hidden, num_outputs, requires_grad=True) * 0.01)
b2 = nn.Parameter(torch.zeros(num_outputs, requires_grad=True))

params = [W1, b1, W2, b2]

activation func

def relu(X):
    a = torch.zeros_like(X)
    return torch.max(X, a)

model

def net(X):
    X = X.reshape((-1, num_inputs))
    H = relu(X@W1 + b1)
    return (H@W2 + b2)

loss

loss = nn.CrossEntropyLoss(reduction='none')

train

num_epochs, lr = 10, 0.1
updater = torch.optim.SGD(params, lr=lr)
d2l.train_ch13(net, train_iter, test_iter, loss, num_epochs, updater)

简洁实现

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

model

net=nn.Sequential(nn.Flatten(),
                  nn.Linear(784,256),
                  nn.ReLU(),
                  nn.Linear(256,10))
def init_weights(m):
    if type(m)==nn.Linear:
        nn.init.normal_(m.weight,std=0.01)

net.apply(init_weights)

args

batch_size,lr,num_epochs=256,0.1,10

loss

loss=nn.CrossEntropyLoss(reduction='none')

optim

trainer=torch.optim.SGD(net.parameters(),lr=lr)

datasets

train_iter,test_iter=d2l.load_data_fashion_mnist(batch_size)

### train

d2l.train_ch13(net=net, train_iter=train_iter, test_iter=test_iter, loss=loss, num_epochs=num_epochs, trainer=trainer)