創(chuàng)建一個網(wǎng)絡(luò)
建立一個簡單的三層的網(wǎng)絡(luò)麦萤。
import torch
from torch import nn
from torch.utils.data import DataLoader
from torchvision import datasets
from torchvision.transforms import ToTensor, Lambda
training_data = datasets.FashionMNIST(
root="data",
train=True,
download=True,
transform=ToTensor()
)
test_data = datasets.FashionMNIST(
root="data",
train=False,
download=True,
transform=ToTensor()
)
train_dataloader = DataLoader(training_data, batch_size=64)
test_dataloader = DataLoader(test_data, batch_size=64)
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()
給出網(wǎng)絡(luò)參數(shù)
learning_rate = 1e-3 #學(xué)習(xí)率杖小，每次沿梯度方向前進(jìn)步長
batch_size = 64 #每次跟新梯度的數(shù)據(jù)數(shù)
epochs = 5 #總訓(xùn)練迭代次數(shù)
損失函數(shù)
常見的順勢函數(shù)為：
nn.MSELoss : Mean Square Error
nn.NLLoss : Negative Log Likelihood
nn.CrossEntropyLoss: nnLogSoftmax + nn.NLLLoss
# Initialize the loss function
loss_fn = nn.CrossEntropyLoss()
優(yōu)化器
optimizer表示網(wǎng)絡(luò)使用何種方法(這里主要涉及梯度和學(xué)習(xí)率)更新網(wǎng)絡(luò)參數(shù)的過程韧掩，常用的梯度更新方式有：
SGD: 隨機(jī)梯度下降法（stochatic gradient decent）
ADAM: 考慮動量的梯度方法
RMSProp:
詳情可參見官方說明: https://pytorch.org/docs/stable/optim.html
一篇非常好的介紹各種梯度方法的博客: https://ruder.io/optimizing-gradient-descent/index.html#adam
optimizer = torch.optim.SGD(model.parameters(), lr=learning_rate)
在每個優(yōu)化循環(huán)中主要包含如下三步：
optimizer.zero_grad(): 清空梯度累計
loss.backward(): 方向傳播遗菠，使用loss函數(shù)累計梯度（累積和batch有關(guān)系）
optimizer.step(): 使用累計的梯度，更新參數(shù)
完整的訓(xùn)練過程
定義一個train_loop()函數(shù)進(jìn)行網(wǎng)絡(luò)訓(xùn)練
定一個test_loop()函數(shù)反饋網(wǎng)絡(luò)預(yù)測結(jié)果
相關(guān)函數(shù)定義如下：
def train_loop(dataloader, model, loss_fn, optimizer):
size = len(dataloader.dataset)
for batch, (X, y) in enumerate(dataloader):
# Compute prediction and loss
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_loop(dataloader, model, loss_fn):
size = len(dataloader.dataset)
num_batches = len(dataloader)
test_loss, correct = 0, 0
with torch.no_grad():
for X, y in dataloader:
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")
一個完整的訓(xùn)練啊犬，優(yōu)化過程：
loss_fn = nn.CrossEntropyLoss()
optimizer = torch.optim.SGD(model.parameters(), lr=learning_rate)
epochs = 10
for t in range(epochs):
print(f"Epoch {t+1}\n-------------------------------")
train_loop(train_dataloader, model, loss_fn, optimizer)
test_loop(test_dataloader, model, loss_fn)
print("Done!")
訓(xùn)練結(jié)果的載入和保存
torch 在網(wǎng)絡(luò)完成訓(xùn)練后虎锚，可以將網(wǎng)絡(luò)參數(shù)以字典（ .state_dict() ）的形式保存為.pth文件。
import torch
import torchvision.models as models
model = models.vgg16(pretrained=True)
torch.save(model.state_dict(), 'model_weights.pth')
同樣也可以講一訓(xùn)練好的網(wǎng)絡(luò)參數(shù)導(dǎo)入：
model = models.vgg16() # we do not specify pretrained=True, i.e. do not load default weights
model.load_state_dict(torch.load('model_weights.pth'))
model.eval()
保存網(wǎng)絡(luò)模型結(jié)構(gòu)
torch.save(model, 'model.pth')
載入網(wǎng)絡(luò)模型結(jié)構(gòu)
model = torch.load('model.pth')
