首先,加载MNIST数据集
import torch
from torchvision import transforms,datasets # datasets自带数据集MNIST
from torch.utils.data import DataLoader
def main():
# 把MNIST数据集加载进来
mnist_train = datasets.MNIST('mnist',True,transform=transforms.Compose([
transforms.ToTensor()
]),download=True)
# 把数据集加载到DataLoader中
mnist_train = DataLoader(mnist_train,batch_size=32,shuffle=True)
# 把MNIST数据集加载进来
mnist_test = datasets.MNIST('mnist',True,transform=transforms.Compose([
transforms.ToTensor()
]),download=True)
# 把数据集加载到DataLoader中
mnist_test = DataLoader(mnist_test,batch_size=32,shuffle=True)
# 构建一个迭代器
x,_ = iter(mnist_train).next() # 不返回label,因为这是无监督学习
print('x:',x.shape) # x:torch.Size([32, 1, 28, 28])
if __name__ == '__main__':
main()
之后写出Auto-Ecoders的网络结构&