Pytorch Tutorial 学习笔记（四）Tensorboard

Pytorch Tutorial 学习笔记（四）Tensorboard

Pytorch官方文档：https://pytorch.org/tutorials/

Tensorboard是pytorch中可视化工具，使用前需要安装pytorch，tensorboard， Matplotlib

安装命令：

安装命令有两种，使用其中一种即可：

使用 conda:

conda install pytorch torchvision -c pytorch` `conda install matplotlib tensorboard

使用 pip:

pip install torch torchvision matplotlib tensorboard

安装完成后需要重新加载环境。

使用

• 导入

# PyTorch 内置函数以及模型使用
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim

# 图像数据操作
import torchvision
import torchvision.transforms as transforms

# Image 展示
import matplotlib.pyplot as plt
import numpy as np

# PyTorch TensorBoard 需要命令
from torch.utils.tensorboard import SummaryWriter

• 可视化图像，使用matplotlib

# 将图像数据转化为tensor类型
transform = transforms.Compose(
    [transforms.ToTensor(),
    transforms.Normalize((0.5,), (0.5,))])

# 将训练集和验证集分开并加载
training_set = torchvision.datasets.FashionMNIST('./data',
    download=True,
    train=True,
    transform=transform)
validation_set = torchvision.datasets.FashionMNIST('./data',
    download=True,
    train=False,
    transform=transform)

training_loader = torch.utils.data.DataLoader(training_set,
                                              batch_size=4,
                                              shuffle=True,
                                              num_workers=2)


validation_loader = torch.utils.data.DataLoader(validation_set,
                                                batch_size=4,
                                                shuffle=False,
                                                num_workers=2)

# 类别标签
classes = ('T-shirt/top', 'Trouser', 'Pullover', 'Dress', 'Coat',
        'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle Boot')

# Helper function for inline image display
def matplotlib_imshow(img, one_channel=False):
    if one_channel:
        img = img.mean(dim=0)
    img = img / 2 + 0.5     # unnormalize
    npimg = img.numpy()
    if one_channel:
        plt.imshow(npimg, cmap="Greys")
    else:
        plt.imshow(np.transpose(npimg, (1, 2, 0)))

# 每次加载展示4张图片
dataiter = iter(training_loader)
images, labels = dataiter.next()

# 创建栅格并展示，展示的是灰度图像
img_grid = torchvision.utils.make_grid(images)
matplotlib_imshow(img_grid, one_channel=True)

• 可视化图像，使用tensorboard

# 默认导出的log_dir文件是 "runs"，但是可以自己定义
# torch.utils.tensorboard.SummaryWriter is imported above
writer = SummaryWriter('runs/fashion_mnist_experiment_1')

# 将图片数据加载到TensorBoard log dir
writer.add_image('Four Fashion-MNIST Images', img_grid)
writer.flush()

# 在pycharm终端输入:
#   tensorboard --logdir=runs
# 在浏览器中打开网址 http://localhost:6006/

• 可视化训练过程

class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.conv1 = nn.Conv2d(1, 6, 5)
        self.pool = nn.MaxPool2d(2, 2)
        self.conv2 = nn.Conv2d(6, 16, 5)
        self.fc1 = nn.Linear(16 * 4 * 4, 120)
        self.fc2 = nn.Linear(120, 84)
        self.fc3 = nn.Linear(84, 10)

    def forward(self, x):
        x = self.pool(F.relu(self.conv1(x)))
        x = self.pool(F.relu(self.conv2(x)))
        x = x.view(-1, 16 * 4 * 4)
        x = F.relu(self.fc1(x))
        x = F.relu(self.fc2(x))
        x = self.fc3(x)
        return x


net = Net()
criterion = nn.CrossEntropyLoss()
optimizer = optim.SGD(net.parameters(), lr=0.001, momentum=0.9)


print(len(validation_loader))
for epoch in range(1):  # 训练的轮数
    running_loss = 0.0

    for i, data in enumerate(training_loader, 0):
        # 在一个epoch中训练
        inputs, labels = data
        optimizer.zero_grad()
        outputs = net(inputs)
        loss = criterion(outputs, labels)
        loss.backward()
        optimizer.step()

        running_loss += loss.item()
        if i % 1000 == 999:    # Every 1000 mini-batches...
            print('Batch {}'.format(i + 1))
            # 进行验证集测试，检查训练结果
            running_vloss = 0.0

            net.train(False) # 关闭训练模式，也就是不反馈梯度
            for j, vdata in enumerate(validation_loader, 0):
                vinputs, vlabels = vdata
                voutputs = net(vinputs)
                vloss = criterion(voutputs, vlabels)
                running_vloss += vloss.item()
            net.train(True) # 打开训练模式

            avg_loss = running_loss / 1000
            avg_vloss = running_vloss / len(validation_loader)

            # 将损失数据加载到Tensorboard中进行查看
            writer.add_scalars('Training vs. Validation Loss',
                            { 'Training' : avg_loss, 'Validation' : avg_vloss },
                            epoch * len(training_loader) + i)

            running_loss = 0.0
print('Finished Training')

writer.flush()

• 可视乎模型

# 加载一个小批量数据
dataiter = iter(training_loader)
images, labels = dataiter.next()

# add_graph() 将会跟踪输入图片轨迹，并且将其转化成一个可视化图表
writer.add_graph(net, images)
writer.flush()

• 加载函数

from torch.utils.tensorboard import SummaryWriter
writer = SummaryWriter()
r = 5
for i in range(100):
    writer.add_scalars('run_14h', {'xsinx':i*np.sin(i/r),
                                    'xcosx':i*np.cos(i/r),
                                    'tanx': np.tan(i/r)}, i)
writer.close()