对于神经网络中的内容

self.conv1 如果将输出维度调整为16，那么不能reshape成3维度，但是可以reshape成8维度，这部分暂时不能理解，如果可以解决，那么就可以随意调整神经网络的过程。

最后还是将卷积的输出维度调整到了3，因为维度为3才能将数据集在tensorboard中展示




import torch
import torchvision
from torch.nn import Conv2d
from torch.nn import MaxPool2d
from torch import nn
from torch.utils.data import DataLoader
from torch.utils.tensorboard import SummaryWriter

dataset = torchvision.datasets.CIFAR10("./data", train=False,
transform=torchvision.transforms.ToTensor(), download=True)
dataloader = DataLoader(dataset, batch_size=4)

for data in dataloader:
    imgs, targets = data
    print(imgs.shape)


# 自己编写一个神经网络
class DSC(nn.Module):
    def __init__(self) -> None:
        super().__init__()
        self.conv1 = Conv2d(in_channels=3, out_channels=3, kernel_size=3, stride=1, padding=2) # 卷积层
        self.conv2 = Conv2d(in_channels=16, out_channels=32, kernel_size=3, stride=1, padding=1)
        self.conv3 = Conv2d(in_channels=32, out_channels=64, kernel_size=3, stride=2, padding=0)  
        self.pool1 = MaxPool2d(kernel_size=5, ceil_mode=True) # 池化层 上取整
        self.pool2 = MaxPool2d(kernel_size=3, ceil_mode=True)
        self.relu = nn.ReLU()

    def forward(self,x):
        x = self.relu(self.conv1(x))
        # x = self.pool1(x)
        # for i in range(2):
        #     x = self.relu(self.conv2(x))
        # x = self.pool2(x)
        # for i in range(2):
        #     x = self.relu(self.conv3(x))
        return x

dsc = DSC()
writer = SummaryWriter("test_logs")

for data in dataloader:
    imgs, targets = data
    print(imgs.shape)

step = 0
for data in dataloader:
    imgs, targets = data
    output = dsc(imgs)
    print(output.shape)
    writer.add_images("imgs", imgs, step)
    # output = torch.reshape(output, (-1, 8, 34, 34))
    # output = torch.reshape(output, (-1, 3, 30, 30))
    writer.add_images("output", output, step)
    step = step + 1