PyTorch深度学习快速入门教程【小土堆】 学习笔记_小土堆深度学习

学习视频地址 —— B站

PyTorch深度学习快速入门教程（绝对通俗易懂！）【小土堆】

1 pyTorch安装和配置

彻底卸载anaconda

conda install anaconda-clean
anaconda-clean --yes
再用控制面板
1
2
3

anaconda 卸载环境 ：conda uninstall -n yyy --all

• anaconda 安装路径：D:\anaconda3
• 创建环境： conda create -n pytorch python=3.9
• 
• 切换环境 ： conda activate pytorch
• 查看目前已经安装的工具包：pip list

Q 安装pytorch?

1. 进入pytorch首页 下拉，https://pytorch.org/
2. 查看gpu型号： 任务管理器-性能 - gpu
3. 
4. 首先更改下载源，，然后，使用官网上的指令安装，直接安装torch一直下载不下来

已经测试，速度很慢，30kB/s

 conda create -n pytorch python=3.9
conda activate pytorch

conda config --remove-key channels  
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/r
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/msys2
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud

版权声明：本文为CSDN博主「StarryHuangx」的原创文章，遵循CC 4.0 BY-SA版权协议，转载请附上原文出处链接及本声明。
原文链接：https://blog.csdn.net/weixin_46017950/article/details/123280653

pip3 install torch torchvision torchaudio
1
2
3
4
5
6
7
8
9
10
11
12
13

已经测试，最有效的办法，速度最快,3MB/s

conda create -n pytorch python=3.9
conda activate pytorch


conda config --set show_channel_urls yes
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/ 


#使用的这个 —— 未成功
conda install pytorch==1.9.1 torchvision==0.10.1 torchaudio==0.9.1 -c pytorch

# 用的这——未成功
pip install torch==1.10.1+cu111 torchvision==0.11.2+cu111 torchaudio==0.10.1 -f https://download.pytorch.org/whl/torch_stable.html


# 这个 未成功
pip install torch==1.10.1+cu111 torchvision==0.11.2+cu111 torchaudio==0.10.1 -f https://download.pytorch.org/whl/cu111/torch_stable.html

pip install torch==1.10.1+cu102 torchvision==0.11.2+cu102 torchaudio==0.10.1 -f https://download.pytorch.org/whl/cu102/torch_stable.html


# 这个可以,但是gpu太low, cuda 版本太高，，gpu上无法训练
pip install torch==1.10.2+cu113 torchvision==0.11.3+cu113 torchaudio===0.10.2+cu113 -f https://download.pytorch.org/whl/cu113/torch_stable.html  

# 11.1
pip install torch==1.10.1+cu111 torchvision==0.11.2+cu111 torchaudio==0.10.1 -f https://download.pytorch.org/whl/torch_stable.html

pip install torch==1.10.1+cu111 torchvision==0.11.2+cu111 torchaudio==0.10.1 -f https://download.pytorch.org/whl/cu111/torch_stable.html

作者：insightya https://www.bilibili.com/read/cv15186754/ 出处：bilibili
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33

2 python 编辑器的安装

pyCharm 安装注意事项

• 在 Installation Options 中，，勾选 .py

pyCharm 配置

1. 打开
2. 新建项目 的过程中
3. 选择 Existing interpreter
4. 手动找到编译器，， 选择 conda Environment ->
5. 输入： D:/anaconda/envs/pytroch/python.exe

检测是否正常运行

1. 点 “” python console “”

import torch
torch.cuda.is_available()
1
2

默认 Jupiter 安装在 base 环境中，，所以，现在 要在 新环境中 在安装一遍 jupiter。

新环境中安装jupyter

# open  anaconda命令行
conda activate pytorch
conda list
conda install nb_conda
jupyter notebook
1
2
3
4
5

jupyter 使用指南

import torch
torch.cuda.is_available()
1
2

shift + 回车： 表示 创建一个新的代码块，并且运行上一个代码块

3 为什么 返回 false

我的是true ，跳过

4 python 学习中的两个 法宝函数

dir() : 打开
help() : 说明书

dir(pytorch)
1

5 pyCharm 和 Jupyter使用及对比

6 pytorch 加载数据

Dataset
提供 获取数据及其lable的方法
能得到 总共有 多少数据
能获取 每一个数据 及其 label

Dataloader
把 数据打包，，方便 后续使用

代开 Jupyter

conda activate pytorch
jupyter notebook
1
2

7 Dataset 实战

安装opencv

conda activate pytorch
conda install opencv-python
1
2

from torch.utils.data import Dataset
from PIL import Image
import os


class MyData(Dataset):

    def __init__(self, root_dir, label_dir):
        self.root_dir = root_dir
        self.label_dir = label_dir
        self.path = os.path.join(self.root_dir, self.label_dir)
        self.img_path = os.listdir(self.path)

    def __getitem__(self, idx):
        img_name = self.img_path[idx]
        img_item_path = os.path.join( self.root_dir, self.label_dir, img_name )
        img = Image.open(img_item_path)
        label = self.label_dir
        return img, label

    def __len__(self):
        return len(self.img_path)

root_dir = "hymenoptera_data/hymenoptera_data/train"
ants_label_dir = "ants"
bees_label_dir = "bees"
ants_dataset = MyData(root_dir, ants_label_dir)
bees_dataset = MyData(root_dir, bees_label_dir)

train_dataset = ants_dataset + bees_dataset
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30

8 Tensorboard 使用 数据可视化

SummaryWriter使用

1. 安装tensorboard

pip install tensorboard
1

2. 报错
   
   包不匹配,解决办法如下

 pip install setuptools==59.5.0
conda install tensorboard
1
2

运行结果是，，多了一个Logs文件夹

数据可视化：
一定要切换到Logs的上一层目录中

conda activate pytorch

D:
cd D:\pycharm-workplace\pythonProject

# 切换到 Logs的上一层目录中
tensorboard --logdir=logs
1
2
3
4
5
6
7

9 tensorboard使用2

利用opencv读取文件，转化为Numpy数组

pip install opencv-python
1

from torch.utils.tensorboard import SummaryWriter
import numpy as np
from PIL import Image

writer = SummaryWriter("logs")
image_path = "hymenoptera_data/hymenoptera_data/train/ants/0013035.jpg"
img_PIL = Image.open(image_path)
img_array = np.array(img_PIL)


writer.add_image("lyy", img_array, 1, dataformats='HWC')

for i in range(100):
    writer.add_scalar("y=2x", 3*i, i)

writer.close()

print("end")
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

10 Transfroms 使用1——主要是对图片进行变换

ToTensor（）使用—— 把 imgge 或 numpy 转换为 tensor

from PIL import Image
from torchvision import transforms

img_path = "hymenoptera_data/hymenoptera_data/train/ants/0013035.jpg"
img = Image.open(img_path)

# tensor_trans 是一个类
tensor_trans = transforms.ToTensor() # 吧 imgge 或 numpy 转换为 tensor
tensor_img = tensor_trans(img)

print(tensor_img)

1
2
3
4
5
6
7
8
9
10
11
12

11 Transfroms 使用2——主要是对图片进行变换

from PIL import Image
from torch.utils.tensorboard import SummaryWriter
from torchvision import transforms

img_path = "hymenoptera_data/hymenoptera_data/train/ants/0013035.jpg"
img = Image.open(img_path)


writer = SummaryWriter("logs")


tensor_trans = transforms.ToTensor()
tensor_img = tensor_trans(img)

# print(tensor_img)

writer.add_image("lyy", tensor_img)
writer.close()


1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

12 常见transforms使用

normalize

from PIL import Image
from torch.utils.tensorboard import SummaryWriter
from torchvision import transforms

img_path = "hymenoptera_data/16.jpg"
img = Image.open(img_path)

writer = SummaryWriter("logs")

# totensor使用
tensor_trans = transforms.ToTensor()
tensor_img = tensor_trans(img)
writer.add_image("lyy", tensor_img)

# Normalize 使用
trans_norm = transforms.Normalize([111,111,111],[10,10,10])
img_norm = trans_norm(tensor_img)
writer.add_image("normalize", img_norm, 2)

writer.close()

print("end")
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22

13 常见transforms使用2

注意事项

关注输入输出，看官方文档，关注方法的参数

from PIL import Image
from torch.utils.tensorboard import SummaryWriter
from torchvision import transforms

img_path = "hymenoptera_data/16.jpg"
img = Image.open(img_path)


writer = SummaryWriter("logs")


# totensor使用
tensor_trans = transforms.ToTensor()
tensor_img = tensor_trans(img)
writer.add_image("lyy", tensor_img)

# Normalize 使用
trans_norm = transforms.Normalize([111,111,111],[10,10,10])
img_norm = trans_norm(tensor_img)
writer.add_image("normalize", img_norm, 2)

# resize
trans_resize = transforms.Resize( (512, 512))
img_resize = trans_resize(img) # 输入的是Image类型的图像
img_resize_tensor = transforms.ToTensor()

tensor_img = tensor_trans(img_resize)
writer.add_image("resize", tensor_img, 0)

# compose 用法
trans_resize_2 =transforms.Resize(123)
trans_compose = transforms.Compose([trans_resize_2, tensor_trans])
img_resize_2 = trans_compose(img)
writer.add_image("resize2", img_resize_2,1)

# randomCrop
trans_random = transforms.RandomCrop((20,50))
trans_compose_2 = transforms.Compose([trans_random, tensor_trans])
for i in range(10):
    img_crop = trans_compose_2(img)
    writer.add_image("randomCrop", img_crop, i)


writer.close()

print("end")
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46

14 torchvision数据集使用

下载网上的数据集，及使用

import torchvision
from torch.utils.tensorboard import SummaryWriter

dataset_transfrom = torchvision.transforms.Compose([
    torchvision.transforms.ToTensor()
])

train_set = torchvision.datasets.CIFAR10(root="./cifar10", train=True, transform=dataset_transfrom, download=True)
test_set = torchvision.datasets.CIFAR10(root="./cifar10", train=False, transform=dataset_transfrom, download=True)

writer = SummaryWriter("p10")
for i in range(10):
    img, target = test_set[i]
    writer.add_image("test_set", img, i)

writer.close()
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

DataLoader 使用

Dataset 有数据，和标签
DataLoader，吧数据 加载 到神经网络

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

test_data = torchvision.datasets.CIFAR10("./cifar10", train=False, transform=torchvision.transforms.ToTensor())
test_loader = DataLoader(dataset=test_data, batch_size=64, shuffle=True, num_workers=0, drop_last=False)

writer = SummaryWriter("dataloader")

step = 0
for data in test_loader:
    imgs, targets = data
    writer.add_images("dataloader", imgs, step)
    step = step + 1

writer.close()\
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

17 卷积操作

import torch
import torch.nn.functional as F

input = torch.tensor([[1, 2, 0, 3, 1],
                      [0, 1, 2, 3, 1],
                      [1, 2, 1, 0, 0],
                      [5, 2, 3, 1, 1],
                      [2, 1, 0, 1, 1]])
kernal = torch.tensor([[1, 2, 1],
                       [0, 1, 0],
                       [2, 1, 0]])

input = torch.reshape(input, (1,1,5,5))
kernal = torch.reshape(kernal, (1,1,3,3))

print(input.shape)
print(kernal.shape)

output = F.conv2d(input, kernal, stride=2, padding=1)
print(output)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

18 神经网络 —— 卷积层

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

dataset = torchvision.datasets.CIFAR10("./cifar10",train=False,transform=torchvision.transforms.ToTensor(), download=False)
dataloader = DataLoader(dataset, batch_size=64)

class Lyy(nn.Module):
    def __init__(self):
        super(Lyy, self).__init__()
        self.conv1 = Conv2d(3, 6, 3, 1)

    def forward(self, x):
        x = self.conv1(x)
        return x

lyy = Lyy()

writer = SummaryWriter("logs")
step = 0
for data in dataloader:
    imgs, targets = data
    output = lyy(imgs)
    writer.add_images("input", imgs, step)

    output = torch.reshape(output, (-1,3,30,30))
    writer.add_images("output", output, step)

    step += 1

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33

19 最大池化层

移动步长，默认是 卷积核的大小

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

dataset = torchvision.datasets.CIFAR10('./cifar10', train=False, transform=torchvision.transforms.ToTensor(),download=False)
dataloader = DataLoader(dataset, 64, False)


input = torch.tensor([[1, 2, 0, 3, 1],
                      [0, 1, 2, 3, 1],
                      [1, 2, 1, 0, 0],
                      [5, 2, 3, 1, 1],
                      [2, 1, 0, 1, 1]], dtype=torch.float32)
kernal = torch.tensor([[1, 2, 1],
                       [0, 1, 0],
                       [2, 1, 0]])

input = torch.reshape(input, (1,1,5,5))
# kernal = torch.reshape(kernal, (1,1,3,3))

class Lyy(nn.Module):
    def __init__(self):
        super(Lyy, self).__init__()
         # self.conv1 = Conv2d(3, 3, 3, 1,padding=1)
        self.maxpool =MaxPool2d(kernel_size=3, ceil_mode=True)

    def forward(self, x):
        # x = self.conv1(x)
        x = self.maxpool(x)
        return x

lyy = Lyy()
# output = lyy(input)

writer = SummaryWriter("logs")
step = 0
for data in dataloader:
    imgs, tagerts = data
    writer.add_images("input", imgs, step)
    output = lyy(imgs)
    writer.add_images("output", output, step)
    step += 1

writer.close()
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47

20 非线性激活

import torch
import torchvision.datasets
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, 2, 0, 3, 1],
                      [0, 1, 2, 3, 1],
                      [1, 2, 1, 0, 0],
                      [5, 2, 3, 1, 1],
                      [2, 1, 0, 1, 1]], dtype=torch.float32)

input = torch.reshape(input, (-1,1, 5, 5))


dataset = torchvision.datasets.CIFAR10("./cifar10", False, transform=torchvision.transforms.ToTensor())
dataloader = DataLoader(dataset, 64)


class Lyy(nn.Module):
    def __init__(self):
        super(Lyy, self).__init__()
        self.relu1 = ReLU()
        self.sigmoid = Sigmoid()

    def forward(self, input):
        return self.sigmoid(input)

lyy = Lyy()


step = 0
writer = SummaryWriter("logs")

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

writer.close()

print("end")
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45

21 线性层 和 其它层

为正确运行

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

dataset = torchvision.datasets.CIFAR10('./cifar10', False, transform=torchvision.transforms.ToTensor())
dataloader = DataLoader(dataset, batch_size=64)


class Lyy(nn.Module):
    def __init__(self):
        super(Lyy, self).__init__()
        super.linear1 = Linear(196608,10)

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

lyy = Lyy()

write = SummaryWriter("logs")

step = 0
for data in dataloader:
    imgs, targers = data
    print(imgs.shape)
    write.add_images("input", imgs, step)
    output = torch.reshape(imgs, (1,1,1, -1))
    print(output.shape)

    write.add_images("output", output, step)
write.close()
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34

22 神经网络搭建小实战

计算 Padding 和 Stride 是多少？

Dilation 默认等于 1

RuntimeError: mat1 and mat2 shapes cannot be multiplied (64x1024 and 10240x64) 原因

原因： 网络参数设置不合适！！

from torch import nn
from torch.nn import Conv2d, MaxPool2d, Flatten, Linear, Sequential
import torch

class Lyy(nn.Module):
    def __init__(self):
        super(Lyy, self).__init__()
        self.conv1 = Conv2d(3, 32, 5, padding=2)
        self.maxpool1 = MaxPool2d(2)
        self.conv2 = Conv2d(32, 32, 5, padding=2)
        self.maxpool2 = MaxPool2d(2)
        self.conv3 = Conv2d(32, 64, 5, padding=2)
        self.maxpool3 = MaxPool2d(2)
        self.flatten = Flatten()
        self.linear1 = Linear(1024, 64)
        self.linear2 = Linear(64, 10)

    def forward(self, x):
        x = self.conv1(x)
        x = self.maxpool1(x)
        x = self.conv2(x)
        x = self.maxpool2(x)
        x = self.conv3(x)
        x = self.maxpool3(x)
        x = self.flatten(x)
        x = self.linear1(x)
        x = self.linear2(x)
        return x

lyy = Lyy()
input = torch.ones((64,3,32,32))
print(input.shape)
output = lyy(input)
print(output.shape)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34

上述代码的，等价替换版本

from torch import nn
from torch.nn import Conv2d, MaxPool2d, Flatten, Linear, Sequential
import torch

class Lyy(nn.Module):
    def __init__(self):
        super(Lyy, self).__init__()
        self.model1 = Sequential(
            Conv2d(3, 32, 5, padding=2),
            MaxPool2d(2),
            Conv2d(32, 32, 5, padding=2),
            MaxPool2d(2),
            Conv2d(32, 64, 5, padding=2),
            MaxPool2d(2),
            Flatten(),
            Linear(1024, 64),
            Linear(64, 10)
        )

    def forward(self, x):
        x = self.model1(x)
        return x

lyy = Lyy()
input = torch.ones((64,3,32,32))
print(input.shape)
output = lyy(input)
print(output.shape)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28

可以用tensorboard绘制流程图

writer = SummaryWriter('logs')
writer.add_graph(lyy, input)
writer.close()
1
2
3

23 损失函数与反向传播

分类问题经常用 交叉熵损失函数

import torch
from torch.nn import L1Loss
from torch import nn

inputs = torch.tensor([1,2,3], dtype=float)
outputs = torch.tensor([1,2,5], dtype=float)

inputs = torch.reshape(inputs, [1,1,1,3])
outputs = torch.reshape(outputs, (1,1,1,3))

loss = L1Loss(reduction='sum')
result = loss(inputs, outputs)
print(result)


loss_mse = nn.MSELoss()
result = loss_mse(inputs, outputs)
print(result)


# 交叉熵损失函数
x = torch.tensor([0.1, 0.2, 0.3])
y = torch.tensor([1])
x = torch.reshape(x, (1, 3))
loss_cross = nn.CrossEntropyLoss()
print(loss_cross)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

import torchvision
from torch import nn
from torch.nn import Sequential, Conv2d, MaxPool2d, Flatten, Linear
from torch.utils.data import DataLoader

dataset = torchvision.datasets.CIFAR10('./cifar10', False, transform=torchvision.transforms.ToTensor())
dataloader = DataLoader(dataset, batch_size=64)

class Lyy(nn.Module):
    def __init__(self):
        super(Lyy, self).__init__()
        self.model1 = Sequential(
            Conv2d(3, 32, 5, padding=2),
            MaxPool2d(2),
            Conv2d(32, 32, 5, padding=2),
            MaxPool2d(2),
            Conv2d(32, 64, 5, padding=2),
            MaxPool2d(2),
            Flatten(),
            Linear(1024, 64),
            Linear(64, 10)
        )

    def forward(self, x):
        x = self.model1(x)
        return x

loss = nn.CrossEntropyLoss()
lyy=Lyy()
for data in dataloader:
    imgs, targets = data
    outputs = lyy(imgs)
    result_loss = loss(outputs, targets)
    result_loss.backward()
    print(result_loss)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35

24 优化器

import torchvision
from torch import nn
from torch.nn import Sequential, Conv2d, MaxPool2d, Flatten, Linear
from torch.utils.data import DataLoader
import torch


dataset = torchvision.datasets.CIFAR10('./cifar10', False, transform=torchvision.transforms.ToTensor())
dataloader = DataLoader(dataset, batch_size=64)


class Lyy(nn.Module):
    def __init__(self):
        super(Lyy, self).__init__()
        self.model1 = Sequential(
            Conv2d(3, 32, 5, padding=2),
            MaxPool2d(2),
            Conv2d(32, 32, 5, padding=2),
            MaxPool2d(2),
            Conv2d(32, 64, 5, padding=2),
            MaxPool2d(2),
            Flatten(),
            Linear(1024, 64),
            Linear(64, 10)
        )

    def forward(self, x):
        x = self.model1(x)
        return x

loss = nn.CrossEntropyLoss()
lyy=Lyy()
optis = torch.optim.SGD(lyy.parameters(), lr=0.1)

for epoch in range(10):
    running_loss = 0
    for data in dataloader:
        imgs, targets = data
        outputs = lyy(imgs)
        result_loss = loss(outputs, targets)

        optis.zero_grad()
        result_loss.backward() # 设置对应的梯度
        optis.step()
        running_loss += result_loss
    print(running_loss)


1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48

25 现有网络模型的使用及修改

AGG模型 + imageNet数据集

1. 安装 scipy包

2. pretrained参数，表示，网络是否训练好

import torchvision.models
from torch import nn
from torch.nn import Conv2d, MaxPool2d, Flatten, Linear, Sequential
import torch

dataset = torchvision.datasets.CIFAR10('./cifar10', False, transform=torchvision.transforms.ToTensor())
#dataloader = DataLoader(dataset, batch_size=64)

vgg16_true = torchvision.models.vgg16(pretrained=True)
vgg16_false = torchvision.models.vgg16(pretrained=False)
# 改造vgg16

print(vgg16_false)
# 更改 vgg16， 增加一层
vgg16_true.classifier.add_module('add_linear', nn.Linear(1000, 10))
print(vgg16_true)

print(vgg16_false)
vgg16_false.classifier[6] = nn.Linear(4096, 10)
print(vgg16_false)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

26 网络模型的保存与读取

方式1 和 方式2 都差不多，推荐 方式2

model_save

import torchvision
import torch

vgg16 = torchvision.models.vgg16(pretrained=False)

# 保存方式1 ： 即保存模型结构，，也保存了参数
torch.save(vgg16, "vgg16_model1.pth")

# 保存方式2 ： 把参数保存成字典，不保存结构 (官方推荐)
torch.save(vgg16.state_dict(), "vgg16_model2.pth")

print("end")
1
2
3
4
5
6
7
8
9
10
11
12

model_load

import torch
import torchvision

# 加载方式1 - 保存方式1
# model = torch.load("vgg16_model1.pth")
# # print(model)


# 加载方式2
vgg16 = torchvision.models.vgg16(pretrained=False)
vgg16.load_state_dict(torch.load("vgg16_model2.pth"))
print(vgg16)
1
2
3
4
5
6
7
8
9
10
11
12

27,28,29 完整的训练 套路

argmax() 针对分类问题很有效

import torch
from torch.utils.tensorboard import SummaryWriter

from model_self import *
import torchvision
from torch.nn import Conv2d
from torch.optim import SGD
from torch.utils.data import DataLoader
from torch import nn
from torch.nn import Sequential, Conv2d, MaxPool2d, Flatten, Linear

# 准备数据及
train_data = torchvision.datasets.CIFAR10('./cifar10', True, transform=torchvision.transforms.ToTensor(),download=False)
test_data = torchvision.datasets.CIFAR10('./cifar10', False, transform=torchvision.transforms.ToTensor(),download=False)


# 求长度
train_data_size = len(train_data)
test_data_size = len(test_data)
print("训练数据及长度:{}".format(train_data_size))
print("测试数据集长度:{}".format(test_data_size))

# 加载数据及
train_dataloader = DataLoader(train_data, batch_size=64)
test_dataloader = DataLoader(test_data, batch_size=64)


# 搭建网络


# 创建网络模型
lyy = Lyy()

# 创建损失函数
loss_fn = nn.CrossEntropyLoss()

# 优化器
learning_rate = 1e-2
optimizer = torch.optim.SGD(lyy.parameters(),lr=learning_rate)


# 设置训练网络参数
total_train_step = 0
total_test_step = 0
epoch = 10


# 添加tensorboard
writer = SummaryWriter("logs")


for i in range(epoch):
    print("-----第{}轮训练开始了-----".format(i+1))

    # 训练步骤开始
    for data in train_dataloader:
        imgs, tragets = data
        output = lyy(imgs)
        loss = loss_fn(output, tragets)

        optimizer.zero_grad()
        loss.backward()
        optimizer.step()

        total_train_step += 1
        if total_train_step % 100 == 0:
            print("训练次数：{}，Loss:{}".format(total_train_step, loss.item()))
            writer.add_scalar("train_loss", loss.item(), total_train_step)

    # 测试步骤开始
    total_test_loss = 0
    total_accuracy = 0

    with torch.no_grad():
        for data in test_dataloader:
            imgs, tragets = data
            output = lyy(imgs)
            loss = loss_fn(output, tragets)
            total_test_loss += loss

            accuracy = (output.argmax(1) -- tragets).sum()
            total_accuracy += accuracy

    print("整体测试机上误差：{}".format(total_test_loss))
    print("整体测试机上的正确率：{}".format(total_accuracy/test_data_size))
    writer.add_scalar("test_loss", total_test_loss, total_test_step)
    writer.add_scalar("test_accuracy", total_accuracy/total_test_step)
    total_test_step += 1

    # torch.save(lyy, "lyy_{}.pth".format(i))
    # print("模型已保存")

writer.close()
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93

30 利用GPU训练

找到 网络模型， 数据， 损失函数，调用.cuda()