softmax回归轻便实现

1.通过深度学习框架的高级API可以大概使实现softmax回归更轻易

2. import torch
3. from torch import nn
4. from d2l import torch as d2l
6. batch_size = 256
7. train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size)

复制代码

2.softmax回归输出层是一个全毗连层

2. net = nn.Sequential(nn.Flatten(), nn.Linear(784, 10))
4. def init_weights(m):
5.     if type(m) == nn.Linear:
6.         nn.init.normal_(m.weight, std=0.01)
8. net.apply(init_weights);

复制代码

3.在交织熵丧失函数中通报为归一化的猜测，盘算softmax及对数

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

复制代码

4.利用学习率为0.1的小批量随机梯度降落作为优化算法

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

复制代码