小土堆Pytorch学习笔记（十二、线性层及其它层介绍）

一、其它层

归一化层（BatchNorm2d)

Reccurent Layers

主要用来文字识别之类的

二、线性层

运行如下代码，查看效果：

import torch
import torchvision
from torch.utils.data import DataLoader

dataset = torchvision.datasets.CIFAR10("./dataset", train=False, transform=torchvision.transforms.ToTensor(),
                                       download=True)
dataloader = DataLoader(dataset, batch_size=64)

for data in dataloader:
    imgs, targets = data
    print(imgs.shape)
    output = torch.reshape(imgs, (1, 1, 1, -1))
    print(output.shape)

结果为：

torch.Size([64, 3, 32, 32])
torch.Size([1, 1, 1, 196608])
torch.Size([64, 3, 32, 32])
torch.Size([1, 1, 1, 196608])

再运行如下代码，把输出图像特征改为10：

import torch
import torchvision
from torch import nn
from torch.nn import Linear
from torch.utils.data import DataLoader

dataset = torchvision.datasets.CIFAR10("./dataset", train=False, transform=torchvision.transforms.ToTensor(),
                                       download=True)
dataloader = DataLoader(dataset, batch_size=64, drop_last=True)

class Cow(nn.Module):

    def __init__(self):
        super(Cow, self).__init__()
        self.linear1 = Linear(196608, 10)

    def forward(self, input):
        output = self.linear1(input)
        return output

cow = Cow()

for data in dataloader:
    imgs, targets = data
    print(imgs.shape)
    output = torch.reshape(imgs, (1, 1, 1, -1))
    print(output.shape)
    output = cow(output)
    print(output.shape)

输出结果：

torch.Size([64, 3, 32, 32])
torch.Size([1, 1, 1, 196608])
torch.Size([1, 1, 1, 10])
torch.Size([64, 3, 32, 32])
torch.Size([1, 1, 1, 196608])
torch.Size([1, 1, 1, 10])

要注意dataloader的drop_last参数要设置为True，将数据集最后16张舍弃，不然会报错，因为最后剩下的16张不满足一个批次64进行线性化。
再换成flattern来做：

import torch
import torchvision
from torch import nn
from torch.nn import Linear
from torch.utils.data import DataLoader

dataset = torchvision.datasets.CIFAR10("./dataset", train=False, transform=torchvision.transforms.ToTensor(),
                                       download=True)
dataloader = DataLoader(dataset, batch_size=64, drop_last=True)

class Cow(nn.Module):

    def __init__(self):
        super(Cow, self).__init__()
        self.linear1 = Linear(196608, 10)

    def forward(self, input):
        output = self.linear1(input)
        return output

cow = Cow()

for data in dataloader:
    imgs, targets = data
    print(imgs.shape)
    output = torch.flatten(imgs)
    print(output.shape)
    output = cow(output)
    print(output.shape)

运行结果如下：

torch.Size([64, 3, 32, 32])
torch.Size([196608])
torch.Size([10])
torch.Size([64, 3, 32, 32])
torch.Size([196608])
torch.Size([10])