1.加载数据集
我们使用 FashionMNIST 数据集。
注：FashionMNIST 数据集 是一个定位在比MNIST图片识别问题稍复杂的数据集,它的设定与MNIST几乎完全一样,包含了 10 类不同类型的衣服、鞋子、包等灰度图片。
以下实例是多分类问题。
- 使用 TorchVision 数据集
# 从开源数据集中下载训练集: 这里使用 FashionMNIST 数据集
training_data = datasets.FashionMNIST(
root="data",
train=True,
download=True,
transform=ToTensor(),
)
# 下载 测试集
test_data = datasets.FashionMNIST(
root="data",
train=False,
download=True,
transform=ToTensor(),
)
- 加载数据集
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(f"Shape of X[N, C, H, W]: {
X.shape}")
print(f"Shape of y: {
y.shape} {
y.dtype}")
break
2. 创建模型
# Get cpu or guv device for training
device = "cuda" if torch.cuda.is_available() else "cpu"
print(f"Using {
device} device")
# define model1
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
pass
model = NeuralNetwork().to(device) # 调用父类函数的 to() 方法
print(model)
3.优化模型参数
# 为了训练一个模型，我们需要一个损失函数 和一个优化器。
loss_fn = nn.CrossEntropyLoss()
optimizer = torch.optim.SGD(model.parameters(), lr=1e-3)
4.训练和测试模型
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}]")
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()
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")
5. 开始训练
epochs = 5 # 迭代次数
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!")
6.保存模型
torch.save(model.state_dict(), "model.pth")
print("Saved PyTorch Model State to model.pth")
7.加载模型
model = NeuralNetwork()
model.load_state_dict(torch.load("model.pth"))
8.预测
classes = [
"T-shirt/top",
"Trouser",
"Pullover",
"Dress",
"Coat",
"Sandal",
"Shirt",
"Sneaker",
"Bag",
"Ankle boot",
]
# 该模型现在可用于进行预测
model.eval()
x, y = test_data[0][0], test_data[0][1]
with torch.no_grad():
pred = model(x)
predicted, actual = classes[pred[0].argmax(0)], classes[y]
print(f'Predicted: "{
predicted}", Actual: "{
actual}"')
运行结果：
Connected to pydev debugger (build 201.7846.77)
Shape of X[N, C, H, W]: torch.Size([64, 1, 28, 28])
Shape of y: torch.Size([64]) torch.int64
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)
)
)
Epoch 1
-------------------------------
loss: 2.302747 [ 0/60000]
loss: 2.296829 [ 6400/60000]
loss: 2.277063 [12800/60000]
loss: 2.274143 [19200/60000]
loss: 2.257148 [25600/60000]
loss: 2.220658 [32000/60000]
loss: 2.236387 [38400/60000]
loss: 2.194870 [44800/60000]
loss: 2.189612 [51200/60000]
loss: 2.161722 [57600/60000]
Test Error:
Accuracy: 48.0%, Avg loss: 2.157671
Epoch 2
-------------------------------
loss: 2.163260 [ 0/60000]
loss: 2.158785 [ 6400/60000]
loss: 2.100264 [12800/60000]
loss: 2.121629 [19200/60000]
loss: 2.077504 [25600/60000]
loss: 2.004310 [32000/60000]
loss: 2.042329 [38400/60000]
loss: 1.955108 [44800/60000]
loss: 1.961550 [51200/60000]
loss: 1.885384 [57600/60000]
Test Error:
Accuracy: 57.4%, Avg loss: 1.891497
Epoch 3
-------------------------------
loss: 1.921233 [ 0/60000]
loss: 1.894850 [ 6400/60000]
loss: 1.776564 [12800/60000]
loss: 1.820165 [19200/60000]
loss: 1.725456 [25600/60000]
loss: 1.653135 [32000/60000]
loss: 1.686808 [38400/60000]
loss: 1.579594 [44800/60000]
loss: 1.608693 [51200/60000]
loss: 1.490921 [57600/60000]
Test Error:
Accuracy: 60.7%, Avg loss: 1.519821
Epoch 4
-------------------------------
loss: 1.587000 [ 0/60000]
loss: 1.551294 [ 6400/60000]
loss: 1.401518 [12800/60000]
loss: 1.474207 [19200/60000]
loss: 1.372276 [25600/60000]
loss: 1.343572 [32000/60000]
loss: 1.372158 [38400/60000]
loss: 1.289430 [44800/60000]
loss: 1.328003 [51200/60000]
loss: 1.217984 [57600/60000]
Test Error:
Accuracy: 63.6%, Avg loss: 1.252232
Epoch 5
-------------------------------
loss: 1.331331 [ 0/60000]
loss: 1.308383 [ 6400/60000]
loss: 1.145954 [12800/60000]
loss: 1.251308 [19200/60000]
loss: 1.140415 [25600/60000]
loss: 1.146848 [32000/60000]
loss: 1.182463 [38400/60000]
loss: 1.111923 [44800/60000]
loss: 1.152718 [51200/60000]
loss: 1.062558 [57600/60000]
Test Error:
Accuracy: 64.9%, Avg loss: 1.087809
Done!
Saved PyTorch Model State to model.pth
Process finished with exit code -1