PyTorch中可视化工具的使用

 import torch
 import torch.nn as nn
 
 class ConvNet(nn.Module):
     def __init__(self):
         super(ConvNet, self).__init__()
 
         self.conv1 = nn.Sequential(
             nn.Conv2d(1, 16, 3, 1, 1),
             nn.ReLU(),
             nn.AvgPool2d(2, 2)
         )
 
         self.conv2 = nn.Sequential(
             nn.Conv2d(16, 32, 3, 1, 1),
             nn.ReLU(),
             nn.MaxPool2d(2, 2)
         )
 
         self.fc = nn.Sequential(
             nn.Linear(32 * 7 * 7, 128),
             nn.ReLU(),
             nn.Linear(128, 64),
             nn.ReLU()
         )
 
         self.out = nn.Linear(64, 10)
 
     def forward(self, x):
         x = self.conv1(x)
         x = self.conv2(x)
         x = x.view(x.size(0), -1)
         x = self.fc(x)
         output = self.out(x)
         return output

MyConvNet = ConvNet()
print(MyConvNet)

pip install hiddenlayer

 import hiddenlayer as h
 vis_graph = h.build_graph(MyConvNet, torch.zeros([1 ,1, 28, 28]))   # 获取绘制图像的对象
 vis_graph.theme = h.graph.THEMES["blue"].copy()     # 指定主题颜色
 vis_graph.save("./demo1.png")   # 保存图像的路径

pip install torchviz

 from torchviz import make_dot
 x = torch.randn(1, 1, 28, 28).requires_grad_(True)  # 定义一个网络的输入值
 y = MyConvNet(x)    # 获取网络的预测值
 
 MyConvNetVis = make_dot(y, params=dict(list(MyConvNet.named_parameters()) + [('x', x)]))
 MyConvNetVis.fORMat = "png"
 # 指定文件生成的文件夹
 MyConvNetVis.directory = "data"
 # 生成文件
 MyConvNetVis.view()

import torchvision
 import torch.utils.data as Data
 # 准备训练用的MNIST数据集
 train_data = torchvision.datasets.MNIST(
     root = "./data/MNIST",  # 提取数据的路径
     train=True, # 使用MNIST内的训练数据
     transform=torchvision.transforms.ToTensor(),    # 转换成torch.tensor
     download=False   # 如果是第一次运行的话，置为True，表示下载数据集到root目录
 )
 
 # 定义loader
 train_loader = Data.DataLoader(
     dataset=train_data,
     batch_size=128,
     shuffle=True,
     num_workers=0
 )
 
 test_data = torchvision.datasets.MNIST(
     root="./data/MNIST",
     train=False,    # 使用测试数据
     download=False
 )
 
 # 将测试数据压缩到0-1
 test_data_x = test_data.data.type(torch.FloatTensor) / 255.0
 test_data_x = torch.unsqueeze(test_data_x, dim=1)
 test_data_y = test_data.targets
 
 # 打印一下测试数据和训练数据的shape
 print("test_data_x.shape:", test_data_x.shape)
 print("test_data_y.shape:", test_data_y.shape)
 
 for x, y in train_loader:
     print(x.shape)
     print(y.shape)
     break

 pip install tensorboardX
 pip install tensorboard

 from tensorboardX import SummaryWriter
 logger = SummaryWriter(log_dir="data/log")
 
 # 获取优化器和损失函数
 optimizer = torch.optim.Adam(MyConvNet.parameters(), lr=3e-4)
 loss_func = nn.CrossEntropyLoss()
 log_step_interval = 100      # 记录的步数间隔
 
 for epoch in range(5):
     print("epoch:", epoch)
     # 每一轮都遍历一遍数据加载器
     for step, (x, y) in enumerate(train_loader):
         # 前向计算->计算损失函数->(从损失函数)反向传播->更新网络
         predict = MyConvNet(x)
         loss = loss_func(predict, y)
         optimizer.zero_grad()   # 清空梯度（可以不写）
         loss.backward()     # 反向传播计算梯度
         optimizer.step()    # 更新网络
         global_iter_num = epoch * len(train_loader) + step + 1  # 计算当前是从训练开始时的第几步(全局迭代次数)
         if global_iter_num % log_step_interval == 0:
             # 控制台输出一下
             print("global_step:{}, loss:{:.2}".format(global_iter_num, loss.item()))
             # 添加的第一条日志：损失函数-全局迭代次数
             logger.add_Scalar("train loss", loss.item() ,global_step=global_iter_num)
             # 在测试集上预测并计算正确率
             test_predict = MyConvNet(test_data_x)
             _, predict_idx = torch.max(test_predict, 1)     # 计算softmax后的最大值的索引，即预测结果
             acc = accuracy_score(test_data_y, predict_idx)
             # 添加第二条日志：正确率-全局迭代次数
             logger.add_scalar("test accuary", acc.item(), global_step=global_iter_num)
             # 添加第三条日志：这个batch下的128张图像
             img = vutils.make_grid(x, nrow=12)
             logger.add_image("train image sample", img, global_step=global_iter_num)
             # 添加第三条日志：网络中的参数分布直方图
             for name, param in MyConvNet.named_parameters():
                 logger.add_histogram(name, param.data.numpy(), global_step=global_iter_num)

import hiddenlayer as hl
 import time
 
 # 记录训练过程的指标
 history = hl.History()
 # 使用canvas进行可视化
 canvas = hl.Canvas()
 
 # 获取优化器和损失函数
 optimizer = torch.optim.Adam(MyConvNet.parameters(), lr=3e-4)
 loss_func = nn.CrossEntropyLoss()
 log_step_interval = 100      # 记录的步数间隔
 
 for epoch in range(5):
     print("epoch:", epoch)
     # 每一轮都遍历一遍数据加载器
     for step, (x, y) in enumerate(train_loader):
         # 前向计算->计算损失函数->(从损失函数)反向传播->更新网络
         predict = MyConvNet(x)
         loss = loss_func(predict, y)
         optimizer.zero_grad()   # 清空梯度（可以不写）
         loss.backward()     # 反向传播计算梯度
         optimizer.step()    # 更新网络
         global_iter_num = epoch * len(train_loader) + step + 1  # 计算当前是从训练开始时的第几步(全局迭代次数)
         if global_iter_num % log_step_interval == 0:
             # 控制台输出一下
             print("global_step:{}, loss:{:.2}".format(global_iter_num, loss.item()))
             # 在测试集上预测并计算正确率
             test_predict = MyConvNet(test_data_x)
             _, predict_idx = torch.max(test_predict, 1)  # 计算softmax后的最大值的索引，即预测结果
             acc = accuracy_score(test_data_y, predict_idx)
 
             # 以epoch和step为索引，创建日志字典
             history.log((epoch, step),
                         train_loss=loss,
                         test_acc=acc,
                         hidden_weight=MyConvNet.fc[2].weight)
 
             # 可视化
             with canvas:
                 canvas.draw_plot(history["train_loss"])
                 canvas.draw_plot(history["test_acc"])
                 canvas.draw_image(history["hidden_weight"])

 from visdom import Visdom
 from sklearn.datasets import  load_iris
 import torch
 import numpy as np
 from PIL import Image

# 绘制图像需要的数据
 iris_x, iris_y = load_iris(return_X_y=True)
 
 # 获取绘图对象，相当于plt
 vis = Visdom()
 
 # 添加折线图
 x = torch.linspace(-6, 6, 100).view([-1, 1])
 sigmoid = torch.nn.Sigmoid()
 sigmoid_y = sigmoid(x)
 tanh = torch.nn.Tanh()
 tanh_y = tanh(x)
 relu = torch.nn.ReLU()
 relu_y = relu(x)
 # 连接三个张量
 plot_x = torch.cat([x, x, x], dim=1)
 plot_y = torch.cat([sigmoid_y, tanh_y, relu_y], dim=1)
 # 绘制线性图
 vis.line(X=plot_x, Y=plot_y, win="line plot", env="main",
          opts={
              "dash" : np.array(["solid", "dash", "dashdot"]),
              "legend" : ["Sigmoid", "Tanh", "ReLU"]
          })

 # 绘制2D和3D散点图
 # 参数Y用来指定点的分布，win指定图像的窗口名称，env指定图像所在的环境，opts通过字典来指定一些样式
 vis.scatter(iris_x[ : , 0 : 2], Y=iris_y+1, win="windows1", env="main")
 vis.scatter(iris_x[ : , 0 : 3], Y=iris_y+1, win="3D scatter", env="main",
             opts={
                 "markersize" : 4,   # 点的大小
                 "xlabel" : "特征1",
                 "ylabel" : "特征2"
             })

 # 添加茎叶图
 x = torch.linspace(-6, 6, 100).view([-1, 1])
 y1 = torch.sin(x)
 y2 = torch.cos(x)
 
 # 连接张量
 plot_x = torch.cat([x, x], dim=1)
 plot_y = torch.cat([y1, y2], dim=1)
 # 绘制茎叶图
 vis.stem(X=plot_x, Y=plot_y, win="stem plot", env="main",
          opts={
              "legend" : ["sin", "cos"],
              "title" : "茎叶图"
          })

 # 计算鸢尾花数据集特征向量的相关系数矩阵
 iris_corr = torch.from_numpy(np.corrcoef(iris_x, rowvar=False))
 # 绘制热力图
 vis.heatmap(iris_corr, win="heatmap", env="main",
             opts={
                 "rownames" : ["x1", "x2", "x3", "x4"],
                 "columnnames" : ["x1", "x2", "x3", "x4"],
                 "title" : "热力图"
             })

 # 可视化图片
 img_Image = Image.open("./example.jpg")
 img_array = np.array(img_Image.convert("L"), dtype=np.float32)
 img_tensor = torch.from_numpy(img_array)
 print(img_tensor.shape)
 
 # 这次env自定义
 vis.image(img_tensor, win="one image", env="MyPlotEnv",
           opts={
               "title" : "一张图像"
           })

 # 可视化文本
 text = "hello world"
 vis.text(text=text, win="text plot", env="MyPlotEnv",
          opts={
              "title" : "可视化文本"
          })