尺寸在线做图网站,wordpress 文章 新窗口,企业微信怎么注册,c2c商城网站开发文章目录 TensorBoard进阶1.设置TensorBoard2.图像数据在TensorBoard中可视化3.模型结构在TensorBoard中可视化#xff08;重点✅#xff09;4.高维数据在TensorBoard中低维可视化5.利用TensorBoard跟踪模型的训练过程#xff08;重点✅#xff09;6.利用TensorBoard给每个… 文章目录 TensorBoard进阶1.设置TensorBoard2.图像数据在TensorBoard中可视化3.模型结构在TensorBoard中可视化重点✅4.高维数据在TensorBoard中低维可视化5.利用TensorBoard跟踪模型的训练过程重点✅6.利用TensorBoard给每个类绘制PR曲线参考 TensorBoard进阶
本教程使用TensorBoard可视化模型、数据和训练过程。展示了如何加载数据通过定义为nn.Module子类的模型将其输入在训练数据上训练该模型并在测试数据上测试它。为了了解发生了什么我们在模型训练时打印出一些统计数据以了解训练是否正在进行。然而我们可以做得更好PyTorch与TensorBoard集成TensorBoard是一种旨在可视化神经网络训练运行结果的工具。本教程使用Fashion-MNIST数据集说明了它的一些功能该数据集可以使用torchvision.datasets读入PyTorch。
引导代码直接运行知道每个功能模块是干什么的就可以了。
# imports
import matplotlib.pyplot as plt
import numpy as npimport torch
import torchvision
import torchvision.transforms as transformsimport torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim# transforms
# 给下面的datasets做准备就是对图像进行处理的将图像数据转换为张量。
transform transforms.Compose([transforms.ToTensor(),transforms.Normalize((0.5,), (0.5,))])# datasets
# 创建并加载FashionMNIST训练集与测试集并对图像数据进行了预处理。
trainset torchvision.datasets.FashionMNIST(./data,downloadTrue,trainTrue,transformtransform)
testset torchvision.datasets.FashionMNIST(./data,downloadTrue,trainFalse,transformtransform)# dataloaders
# 定义了train dataloader与test dataloader用于加载数据的。
trainloader torch.utils.data.DataLoader(trainset, batch_size4,shuffleTrue, num_workers2)testloader torch.utils.data.DataLoader(testset, batch_size4,shuffleFalse, num_workers2)# constant for classes
# 类别标签
classes (T-shirt/top, Trouser, Pullover, Dress, Coat,Sandal, Shirt, Sneaker, Bag, Ankle Boot)# helper function to show an image
# (used in the plot_classes_preds function below)
# 显示图像的函数
def matplotlib_imshow(img, one_channelFalse):if one_channel:img img.mean(dim0)img img / 2 0.5 # unnormalizenpimg img.numpy()if one_channel:plt.imshow(npimg, cmapGreys)else:plt.imshow(np.transpose(npimg, (1, 2, 0)))# 定义了一个Net模型
class Net(nn.Module):def __init__(self):super(Net, self).__init__()self.conv1 nn.Conv2d(1, 6, 5) # 二维卷积self.pool nn.MaxPool2d(2, 2) # 最大池化self.conv2 nn.Conv2d(6, 16, 5) # 二维卷积self.fc1 nn.Linear(16 * 4 * 4, 120) # 全连接self.fc2 nn.Linear(120, 84) # 全连接self.fc3 nn.Linear(84, 10) # 全连接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 xnet Net()criterion nn.CrossEntropyLoss() # 交叉熵损失
optimizer optim.SGD(net.parameters(), lr0.001, momentum0.9) # SGD优化器1.设置TensorBoard
from torch.utils.tensorboard import SummaryWriter# default log_dir is runs - well be more specific here
# 创建SummaryWriter实例
writer SummaryWriter(runs/fashion_mnist_experiment_1)2.图像数据在TensorBoard中可视化
利用add_image(tag, img_tensor, global_stepNone, walltimeNone, dataformatsCHW)函数将图像数据添加到Writer中。
writer.add_image(tag, image_tensor)Examples:
# get some random training images
dataiter iter(trainloader)
images, labels next(dataiter)# create grid of images
img_grid torchvision.utils.make_grid(images)# show images
matplotlib_imshow(img_grid, one_channelTrue)# write to tensorboard
# 一个标签只对应一组图片尽管这段运行多次代码中呈现的是不同的图片组但是写入tensorboard中的图片组仅是第一次运行所保存的图片组。
writer.add_image(four_fashion_mnist_images, img_grid)注意pillow版本应该低于10.0.0否则会报错module pil.image has no attribute antialias。 现在启动TensorBoard注意要写全路径或者在对应的目录下直接tensorboard --logdir runs。Windows下和Linux下不同Windows下路径需要用引号包括Linux不用。
tensorboard --logdir D:\Jupyter\Introduction to Pytorch\runs3.模型结构在TensorBoard中可视化重点✅
利用add_graph(model, input_to_modelNone, verboseFalse, use_strict_traceTrue)函数将模型结构添加到Writer中。
writer.add_graph(model, input_to_model torch.rand(1, 3, 224, 224))Examples:
writer.add_graph(net, images)
writer.close()刷新TensorBoard UI界面会在“Graphs”选项卡看到模型结构。模型结构里面还有维度的变化很方便我们去观察特征的维度变化。 4.高维数据在TensorBoard中低维可视化
利用add_embedding(mat, metadataNone, label_imgNone, global_stepNone, tagdefault, metadata_headerNone)函数将嵌入数据可视化。
writer.add_embedding(features,metadataclass_labels,label_imgimages.unsqueeze(1))Examples
# helper function
def select_n_random(data, labels, n100):Selects n random datapoints and their corresponding labels from a datasetassert len(data) len(labels)perm torch.randperm(len(data))return data[perm][:n], labels[perm][:n]# select random images and their target indices
images, labels select_n_random(trainset.data, trainset.targets)# get the class labels for each image
class_labels [classes[lab] for lab in labels]# log embeddings
features images.view(-1, 28 * 28)
writer.add_embedding(features,metadataclass_labels,label_imgimages.unsqueeze(1))
writer.close()在TensorBoard UI中找到Project选项卡然后就会看到被投影到三维空间的图像数据。 5.利用TensorBoard跟踪模型的训练过程重点✅
利用add_scalar(tag, scalar_value, global_stepNone, walltimeNone, new_styleFalse, double_precisionFalse)函数记录要跟踪的指标来观察模型的训练过程。
writer.add_scalar(loss, loss, epoch)
writer.add_scalar(accuracy, accuracy, epoch)Examples
# helper functionsdef images_to_probs(net, images):Generates predictions and corresponding probabilities from a trainednetwork and a list of imagesoutput net(images)# convert output probabilities to predicted class_, preds_tensor torch.max(output, 1)preds np.squeeze(preds_tensor.numpy())return preds, [F.softmax(el, dim0)[i].item() for i, el in zip(preds, output)]def plot_classes_preds(net, images, labels):Generates matplotlib Figure using a trained network, along with imagesand labels from a batch, that shows the networks top prediction alongwith its probability, alongside the actual label, coloring thisinformation based on whether the prediction was correct or not.Uses the images_to_probs function.preds, probs images_to_probs(net, images)# plot the images in the batch, along with predicted and true labelsfig plt.figure(figsize(12, 48))for idx in np.arange(4):ax fig.add_subplot(1, 4, idx1, xticks[], yticks[])matplotlib_imshow(images[idx], one_channelTrue)ax.set_title({0}, {1:.1f}%\n(label: {2}).format(classes[preds[idx]],probs[idx] * 100.0,classes[labels[idx]]),color(green if preds[idx]labels[idx].item() else red))return figrunning_loss 0.0
for epoch in range(1): # loop over the dataset multiple timesfor i, data in enumerate(trainloader, 0):# get the inputs; data is a list of [inputs, labels]inputs, labels data# zero the parameter gradientsoptimizer.zero_grad()# forward backward optimizeoutputs net(inputs)loss criterion(outputs, labels)loss.backward()optimizer.step()running_loss loss.item()if i % 1000 999: # every 1000 mini-batches...# ...log the running loss# 记录的是每1000个mini-batch所对应的损失变化writer.add_scalar(training loss,running_loss / 1000,epoch * len(trainloader) i)# ...log a Matplotlib Figure showing the models predictions on a random mini-batchwriter.add_figure(predictions vs. actuals,plot_classes_preds(net, inputs, labels),global_stepepoch * len(trainloader) i)running_loss 0.0
print(Finished Training)在TensorBoard UI中的Scalars选项卡查看loss的变化。 6.利用TensorBoard给每个类绘制PR曲线
利用add_pr_curve(tag, labels, predictions, global_stepNone, num_thresholds127, weightsNone, walltimeNone)函数绘制精度召回曲线。
writer.add_pr_curve(classes[class_index],tensorboard_truth,tensorboard_probs)Examples
# 1. gets the probability predictions in a test_size x num_classes Tensor
# 2. gets the preds in a test_size Tensor
# takes ~10 seconds to run
class_probs []
class_label []
with torch.no_grad():for data in testloader:images, labels dataoutput net(images)class_probs_batch [F.softmax(el, dim0) for el in output]class_probs.append(class_probs_batch)class_label.append(labels)test_probs torch.cat([torch.stack(batch) for batch in class_probs])
test_label torch.cat(class_label)# helper function
def add_pr_curve_tensorboard(class_index, test_probs, test_label, global_step0):Takes in a class_index from 0 to 9 and plots the correspondingprecision-recall curvetensorboard_truth test_label class_indextensorboard_probs test_probs[:, class_index]writer.add_pr_curve(classes[class_index],tensorboard_truth,tensorboard_probs,global_stepglobal_step)writer.close()# plot all the pr curves
for i in range(len(classes)):add_pr_curve_tensorboard(i, test_probs, test_label)在TensorBoard UI中的PR Curves选项卡查看每个类的精度-召回曲线Precision-Recall Curve简称 PR 曲线。PR 曲线功能可以用于比较不同分类器或不同模型的性能。通过比较不同模型的 PR 曲线下的面积Area Under the Curve, AUC可以直观地评估哪种模型在特定类别上表现更优对于多分类问题每个类别都有对应的 PR 曲线通过分析每个类别的 PR 曲线可以发现模型在哪些类别上表现较好在哪些类别上存在不足从而针对性地进行改进。 参考
Visualizing Models, Data, and Training with TensorBoard准确率、精确率、召回率、P-R曲线torch.utils.tensorboard
