本节介绍有关机器学习常见任务重的API。请参阅每一节的链接以深入了解。
Working with data
PyTorch有两个有关数据工作的原型：torch.utils.data.DataLoader 和 torch.utils.data.Dataset。Dataset 存储了样本及其对应的标签，而 DataLoader为 Dataset 生成了一个迭代器。
import torch
from torch import nn
from torch.utils.data import DataLoader
from torchvision import datasets
from torchvision.transforms import ToTensor, Lambda, Compose
import matplotlib.pyplot as plt
PyTorch提供了针对特定领域的库，如TorchText，TorchVision和TorchAudio，以上均有datasets。在本教程中，我们将使用一个TorchVision dataset。
torchvision.datasets 模块包含多种真实世界的视觉数据集 Dataset 对象，如CIFAR、COCO（full list here）。本教程中，我们使用FashionMNIST数据集。每个TorchVision Dataset 均包括两个参数：transform 和 target_transform分别用于修改样本和标签。
# 从公开数据集上下载训练数据
training_data = datasets.FashionMNIST(
root='data',
train=True,
download=True,
transform=ToTensor(),
)
# Download test data from open datasets.
test_data = datasets.FashionMNIST(
root='data',
train=False,
download=True,
transform=ToTensor(),
)
输出：
Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/train-images-idx3-ubyte.gz
Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/train-images-idx3-ubyte.gz to data/FashionMNIST/raw/train-images-idx3-ubyte.gz
Extracting data/FashionMNIST/raw/train-images-idx3-ubyte.gz to data/FashionMNIST/raw
Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/train-labels-idx1-ubyte.gz
Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/train-labels-idx1-ubyte.gz to data/FashionMNIST/raw/train-labels-idx1-ubyte.gz
Extracting data/FashionMNIST/raw/train-labels-idx1-ubyte.gz to data/FashionMNIST/raw
Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-images-idx3-ubyte.gz
Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-images-idx3-ubyte.gz to data/FashionMNIST/raw/t10k-images-idx3-ubyte.gz
Extracting data/FashionMNIST/raw/t10k-images-idx3-ubyte.gz to data/FashionMNIST/raw
Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-labels-idx1-ubyte.gz
Downloading http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/t10k-labels-idx1-ubyte.gz to data/FashionMNIST/raw/t10k-labels-idx1-ubyte.gz
Extracting data/FashionMNIST/raw/t10k-labels-idx1-ubyte.gz to data/FashionMNIST/raw
我们将 Dataset 作为参数传递给 DataLoader。这为我们的dataset包装了一个迭代器，并支持自动生成batch、抽样、打乱和多进程数据加载。这里定义了一个大小为64的batch，即，dataloader迭代的每一个元素将返回一个包含64个样本及对应标签的batch。
batch_size = 64
# Create data loaders
train_dataloader = DataLoader(training_data, batch_size=batch_size)
test_dataloader = DataLoader(test_data, batch_size=batch_size)
for X, y in test_dataloader:
print("Shape of X [N, C, H, W]: ", X.shape)
print("Shape of y: ", y.shape, y.dtpe)
break
输出：
Shape of X [N, C, H, W]: torch.Size([64, 1, 28, 28])
Shape of y: torch.Size([64]) torch.int64
Read more abount loading data in PyTorch。
创建模型
为了在PyTorch定义模型，我们创建了一个类，继承自nn.Module。我们在 __init__函数中定义是网络的layers，并在 forward 函数中指定data如何通过网络。为加快神经网络中的操作，若GPU可用，则把其移动到GPU上。
# Get cpu or gpu device for training
device = 'cuda' if torch.cuda.is_availabel() else "cpu"
print(f"Using {device} device")
# Define model
class NeuralNetwork(nn.Module):
def __init__(self):
super(NeuralNetwork, self).__init__()
self.flatten = nn.Flatten()
self.linear_relu_stack = nn.Sequential(
nn.Linear(28*28, 512),
nn.ReLU(),
nn.Linear(512, 512),
nn.ReLU(),
nn.Linear(512, 10)
)
def forward(self, x):
x = self.flatten(x)
logits = self.linear_relu_stack(x)
return logits
model = NeuralNetwork().to(device)
print(model)
输出：
Using cuda device
NeuralNetwork(
(flatten): Flatten(start_dim=1, end_dim=-1)
(linear_relu_stack): Sequential(
(0): Linear(in_features=784, out_features=512, bias=True)
(1): ReLU()
(2): Linear(in_features=512, out_features=512, bias=True)
(3): ReLU()
(4): Linear(in_features=512, out_features=10, bias=True)
)
)
Read more about building neural networks in PyTorch。
优化模型参数
为了训练模型，我们需要一个loss function和一个优化器。
loss_fn = nn.CrossEntropyLoss()
optimizer = torch.optim.SGD(model.parameters(), lr=1e-3)
在单个训练循环中，模型在训练集上作出预测（分批喂给模型），并且反向传播预测误差来调整模型参数。
def train(dataloader, model, loss_fn, optimizer):
size = len(dataloader.dataset)
model.train()
for batch, (X, y) in enumerate(dataloader):
X, y = X.to(device), y.to(device)
# Compute prediction error
pred = model(X)
loss = loss_fn(pred, y)
# Backpropagation
optimizer.zero_grad()
loss.backward()
optimizer.step()
if batch % 100 == 0:
loss, current = loss.item(), batch * len(x)
print(f"loss: {loss:>7f} [{current:>5d}/{size:>5d}]")
我们还可以检查模型在测试集上的性能，确保模型是在学习。
def test(dataloader, model, loss_fn):
size = len(dataloader.dataset)
num_batches = len(dataloader)
model.eval()
test_loss, correct = 0, 0
with torch.no_grad():
for X, y in dataloader:
X, y = X.to(device), y.to(device)
pred = model(X)
test_loss += loss_fn(pred, y).item()
# 第一维度是batch，第二维度是预测值
correct += (pred.argmax(1) == y).type(torch.float).sum().item()
test_loss /= num_batches
correct /= size
print(f"Test Error: \n Accuracy: {(100 * correct):>0.1f}%, Avg loss: {test_loss:>8f}\n")
训练过程由几次迭代（epochs）组成。每一个epoch，模型学习参数，作出更好的预测。我们在每次epoch都打印了模型的准确率和损失，我们希望看到随着每次epoch，准确率升高，而损失降低。
epochs = 5
for t in range(epochs):
print(f"Epoch {t+1}\n-----------------------")
train(train_dataloader, model, loss_fn, optimizer)
test(test_dataloader, model, loss_fn)
print("Done!")
输出：
点击查看代码
Epoch 1
-------------------------------
loss: 2.296330 [ 0/60000]
loss: 2.292454 [ 6400/60000]
loss: 2.276433 [12800/60000]
loss: 2.274833 [19200/60000]
loss: 2.250950 [25600/60000]
loss: 2.225070 [32000/60000]
loss: 2.222952 [38400/60000]
loss: 2.200769 [44800/60000]
loss: 2.202750 [51200/60000]
loss: 2.155594 [57600/60000]
Test Error:
Accuracy: 40.0%, Avg loss: 2.161115
Epoch 2
-------------------------------
loss: 2.169147 [ 0/60000]
loss: 2.165468 [ 6400/60000]
loss: 2.118014 [12800/60000]
loss: 2.129221 [19200/60000]
loss: 2.074899 [25600/60000]
loss: 2.022606 [32000/60000]
loss: 2.033795 [38400/60000]
loss: 1.976709 [44800/60000]
loss: 1.982757 [51200/60000]
loss: 1.881978 [57600/60000]
Test Error:
Accuracy: 57.4%, Avg loss: 1.902724
Epoch 3
-------------------------------
loss: 1.934711 [ 0/60000]
loss: 1.906422 [ 6400/60000]
loss: 1.806563 [12800/60000]
loss: 1.832814 [19200/60000]
loss: 1.717731 [25600/60000]
loss: 1.673628 [32000/60000]
loss: 1.679022 [38400/60000]
loss: 1.602205 [44800/60000]
loss: 1.623030 [51200/60000]
loss: 1.492521 [57600/60000]
Test Error:
Accuracy: 61.5%, Avg loss: 1.529054
Epoch 4
-------------------------------
loss: 1.592845 [ 0/60000]
loss: 1.556097 [ 6400/60000]
loss: 1.417763 [12800/60000]
loss: 1.478243 [19200/60000]
loss: 1.357680 [25600/60000]
loss: 1.356057 [32000/60000]
loss: 1.360733 [38400/60000]
loss: 1.298324 [44800/60000]
loss: 1.329920 [51200/60000]
loss: 1.219030 [57600/60000]
Test Error:
Accuracy: 63.4%, Avg loss: 1.250318
Epoch 5
-------------------------------
loss: 1.324846 [ 0/60000]
loss: 1.306784 [ 6400/60000]
loss: 1.145549 [12800/60000]
loss: 1.245576 [19200/60000]
loss: 1.123671 [25600/60000]
loss: 1.150098 [32000/60000]
loss: 1.164900 [38400/60000]
loss: 1.111517 [44800/60000]
loss: 1.147514 [51200/60000]
loss: 1.059701 [57600/60000]
Test Error:
Accuracy: 64.6%, Avg loss: 1.081113
Done!
Read more about Training your model.
保存模型
保存模型的一个常见方法是序列化内部状态字典（包含模型参数）
torch.save(model.state_dict(), "model.pth")
print("Saved PyTorch Model State to model.pth")
加载模型
加载模型包括重新创建模型结构和把状态字典加载进去两个过程。
model = NeuralNetwork()
model.load_state_dict(torch.load("model.pth"))
模型现在可以用来作出预测了。
classes = [
"T-shirt/top",
"Trouser",
"Pullover",
"Dress",
"Coat",
"Sandal",
"Shirt",
"Sneaker",
"Bag",
"Ankle boot",
]
model.eval()
x, y = test_data[0][0], test_data[0][1]
with torch.no_grad():
pred = model(x)
# 这里batch=1，只想取到值而非tensor，因此，取pred的第一维度，此时，预测值从第二维度变成了第一维度
predicted, actual = classes[pred[0].argmax(0)], classes[y]
print(f'Predicted: "{predicted}", Actual: {actual}')
输出：
Predicted: "Ankle boot", Actual: "Ankle boot"
Read more about Saving & Loading your model.