TensorBoard的使用

1.TensorBoard的使用

1.1 安装TensorBoard

pip install tensorboard

命令行输入 tensorboard --logdir=logs   #logdir = 事件文件所在文件夹名

出现 TensorBoard 2.11.2 at http://localhost:6006/ (Press CTRL+C to quit)

tensorboard --logdir=logs --port=6007 #修改端口，防止端口被占用

 1.2 add_scalar()方法

from torch.utils.tensorboard import SummaryWriter
writer = SummaryWriter("logs")
# writer.add_image()
for i in range(100):
    writer.add_scalar("y=x",i,i)
writer.close()

　运行完之后，会生成tensorboard文件夹，在命令行输入tensorboard --logdir=logs时，此时目录应该在logs的上级目录。

 1.3 add_image()的使用

from torch.utils.tensorboard import SummaryWriter
writer = SummaryWriter("logs")
writer.add_image()

1.4 利用numpy.array(),对PIL图片进行转换

1.5 小练习

 创建一个简单的pytorch模型

import torch
import torch.nn as nn
import numpy as np
from torch.utils.tensorboard import SummaryWriter


# 定义模型
class LinearRegressionModel(nn.Module):
    def __init__(self):
        super(LinearRegressionModel, self).__init__()
        self.linear = nn.Linear(1, 1)  # 输入和输出都是1维

    def forward(self, x):
        return self.linear(x)


# 准备数据
x_train = np.array([[3.3], [4.4], [5.5], [6.71], [6.93], [4.168],
                    [9.779], [6.182], [7.59], [2.167],
                    [7.042], [10.791], [5.313], [7.997], [3.1]], dtype=np.float32)

y_train = np.array([[1.7], [2.76], [2.09], [3.19], [1.694], [1.573],
                    [3.366], [2.596], [2.53], [1.221],
                    [2.827], [3.465], [1.65], [2.904], [1.3]], dtype=np.float32)

x_train = torch.from_numpy(x_train)
y_train = torch.from_numpy(y_train)

# 初始化模型
model = LinearRegressionModel()

# 损失和优化器
criterion = nn.MSELoss()
optimizer = torch.optim.SGD(model.parameters(), lr=0.01)

# 初始化SummaryWriter
writer = SummaryWriter('runs/linear_regression_experiment')

# 训练模型
num_epochs = 100
for epoch in range(num_epochs):
    # 转换为tensor
    inputs = x_train
    targets = y_train

    # 前向传播
    outputs = model(inputs)
    loss = criterion(outputs, targets)

    # 反向传播和优化
    optimizer.zero_grad()
    loss.backward()
    optimizer.step()

    # 记录损失
    writer.add_scalar('Loss/train', loss.item(), epoch)

    if (epoch + 1) % 10 == 0:
        print(f'Epoch [{epoch + 1}/{num_epochs}], Loss: {loss.item():.4f}')

# 关闭SummaryWriter
writer.close()

 观察结果，如图所示：