10 多层感知机
从零实现:
import torch
from torch import nn
from d2l import torch as d2l
#实现relu函数
def relu(X):
a = torch.zeros_like(X)
#就是个max,然后将负数置零
return torch.max(X, a)
#定义模型,两个矩阵乘法和加法即可
def net(X):
X = X.reshape((-1, num_inputs)) #num_inputs = 784,一张图片展平后的长度
H = relu(X@W1 + b1) # 这里“@”代表矩阵乘法
return (H@W2 + b2)
#生成数据集
batch_size = 256
train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size)
#初始化模型参数
num_inputs, num_outputs, num_hiddens = 784, 10, 256
#随机生成w矩阵
W1 = nn.Parameter(torch.randn(
num_inputs, num_hiddens, requires_grad=True) * 0.01)
b1 = nn.Parameter(torch.zeros(num_hiddens, requires_grad=True))
W2 = nn.Parameter(torch.randn(
num_hiddens, num_outputs, requires_grad=True) * 0.01)
b2 = nn.Parameter(torch.zeros(num_outputs, requires_grad=True))
params = [W1, b1, W2, b2]
#定义损失函数
loss = nn.CrossEntropyLoss(reduction='none')
#定义优化算法
lr = 0.1
updater = torch.optim.SGD(params, lr=lr)
#训练模型
num_epochs = 10
d2l.train_ch3(net, train_iter, test_iter, loss, num_epochs, updater)
简单实现:
import torch
from torch import nn
from d2l import torch as d2l
#初始化模型参数的函数
def init_weights(m):
if type(m) == nn.Linear:
nn.init.normal_(m.weight, std=0.01)
#同样需要先将输入展平
net = nn.Sequential(
nn.Flatten(),
nn.Linear(784, 256),
nn.ReLU(),
nn.Linear(256, 10))
#模型初始化
net.apply(init_weights);
#设置参数
batch_size, lr, num_epochs = 256, 0.1, 10
#定义损失函数
loss = nn.CrossEntropyLoss(reduction='none')
#定义优化算法
trainer = torch.optim.SGD(net.parameters(), lr=lr)
#读取数据集
train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size)
#训练模型
d2l.train_ch3(net, train_iter, test_iter, loss, num_epochs, trainer)
PS:简单加多一层:784-1440-256-10