09. Softmax Classifier

import torch
from torchvision import datasets
from torchvision import transforms
from torch.utils.data import DataLoader
import torch.nn.functional as F
import torch.optim as optim

batch_size= 64
transoform =transforms.Compose([
    transforms.ToTensor(),
    transforms.Normalize((0),(1))
])
train_data=datasets.MNIST(root="../",
           train=True,
            transform=transoform,#这里写错了。写成了transforms了
            download = True,
)
train_loader=DataLoader(
    train_data,
    batch_size=32,
    shuffle=True
)

test_data = datasets.MNIST(root="../",
    train=False,
    transform=transoform,
    download=True
)
test_loader=DataLoader(
    dataset=test_data,
    batch_size=32,
    shuffle=False
)


class Mo(torch.nn.Module):
    def __init__(self):
        super(Mo, self).__init__()
        self.l1=torch.nn.Linear(784,12)
        self.l2=torch.nn.Linear(12,10)
    def forward(self,x):
        x=x.view(-1,784)#这里写错了，写成x.view 没写x=
        x=F.relu(self.l1(x))
        x=self.l2(x)
        return x

model=Mo()
criterion=torch.nn.CrossEntropyLoss()
optimizer=optim.SGD(model.parameters(),lr=0.1)

def train(epoch):
    for batch_idx,data in enumerate(train_loader,0):
        x,y=data
        y_pred=model(x)

        optimizer.zero_grad()
        loss =criterion(y_pred,y)
        loss.backward()
        optimizer.step()
        if(batch_idx % 100 ==0):
            print("epoch{},loss={}".format(epoch,loss.item()))
def test():
    correct,total=0,0
    with torch.no_grad():
        for data in enumerate(test_loader):
            x,y=data
            y_pred=model(x)
            _,predict=torch.max(y_pred,dim=1)
            total+=y.size(0)
            correct+=(predict==y).sum().item()

        print("test acc{}".format(correct/total))

for epoch in range(1000):
    train(epoch)
    if(epoch%100==0):
        test()