P19 神经网络-非线性激活

#P19 神经网络-非线性激活

#常用：ReLU、Sigmoid


#nn_relu.py
import torch
import torchvision
from torch import nn
from torch.nn import ReLU, Sigmoid
from torch.utils.data import DataLoader
from torch.utils.tensorboard import SummaryWriter

# input = torch.tensor([[1, -0.5],
#                       [-1, 3]])
#
# input = torch.reshape(input, (-1, 1, 2, 2))#需要有batch_size，因此还是用reshape设置形状，一般设置成(batch_size, channel, H_in, W_in)比较常用
# print(input.shape)

dataset = torchvision.datasets.CIFAR10("./data_nn", train=False, download=True,
                                       transform=torchvision.transforms.ToTensor())

dataloader = DataLoader(dataset, batch_size=64)

class West(nn.Module):
    def __init__(self):
        super(West, self).__init__()
        # self.relu1 = ReLU()#ReLU是使得输入小于0的数为0，大于0的数保持原始值。#inplace=Ture会使得input被output覆盖，即不保留原始值；inplace=False会保留原始值，防止数据丢失。默认为False，因此一般不设置。
        self.sigmoid1 = Sigmoid()#效果是将图加了个白色滤镜

    def forward(self, input):
        # output = self.relu1(input)
        output = self.sigmoid1(input)
        return output

west = West()#将神经网络赋予给west
# output = west(input)
# print(output)

writer = SummaryWriter("./logs_Sigmoid")

step = 0
for data in dataloader:
    imgs, targets = data
    writer.add_images("input", imgs, global_step=step)
    output = west(imgs)
    writer.add_images("output", output, step)
    step += 1

writer.close()


'''
#ReLU附官方网址和说明
ReLU网址
https://pytorch.org/docs/stable/generated/torch.nn.ReLU.html#torch.nn.ReLU
ReLU说明
classtorch.nn.ReLU(inplace=False)
Shape:
Input: (N, ∗)，where means any number of dimensions.
Output: (N, ∗), same shape as the input.

#Sigmoid附官方网址和说明
Sigmoid网址
https://pytorch.org/docs/stable/generated/torch.nn.Sigmoid.html#torch.nn.Sigmoid
Sigmoid说明
classtorch.nn.Sigmoid(*args, **kwargs)
Input: (N, ∗)，where means any number of dimensions.
Output: (N, ∗), same shape as the input.
'''

#ReLU附官方网址和说明
ReLU网址
https://pytorch.org/docs/stable/generated/torch.nn.ReLU.html#torch.nn.ReLU
ReLU说明
classtorch.nn.ReLU(inplace=False)
Shape:
Input: (N, ∗)，where means any number of dimensions.
Output: (N, ∗), same shape as the input.

#Sigmoid附官方网址和说明
Sigmoid网址
https://pytorch.org/docs/stable/generated/torch.nn.Sigmoid.html#torch.nn.Sigmoid
Sigmoid说明
classtorch.nn.Sigmoid(*args, **kwargs)
Input: (N, ∗)，where means any number of dimensions.
Output: (N, ∗), same shape as the input.