预训练和微调代码

数据集：CIFAR10
CIFAR-10数据集由10类32x32的彩色图片组成，一共包含60000张图片，每一类包含6000图片。其中50000张图片作为训练集，10000张图片作为测试集。数据集介绍来自：CIFAR10

图片来源：https://paperswithcode.com/dataset/cifar-10

预训练模型: vgg16

代码

# Imports
import torch
import torchvision
import torch.nn as nn  # All neural network modules, nn.Linear, nn.Conv2d, BatchNorm, Loss functions
import torch.optim as optim  # For all Optimization algorithms, SGD, Adam, etc.
import torch.nn.functional as F  # All functions that don't have any parameters
from torch.utils.data import (
    DataLoader,
)  # Gives easier dataset managment and creates mini batches
import torchvision.datasets as datasets  # Has standard datasets we can import in a nice way
import torchvision.transforms as transforms  # Transformations we can perform on our dataset
from tqdm import tqdm

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

# Hyperparameters
num_classes = 10
learning_rate = 1e-3
batch_size = 1024
num_epochs = 2


class Identity(nn.Module):
    def __init__(self):
        super(Identity, self).__init__()

    def forward(self, x):
        return x

# Load pretrain model & modify it

model = torchvision.models.vgg16(weights='DEFAULT')
# # If you want to do finetuning then set requires_grad = False
# # Remove these two lines if you want to train entire model,
# # and only want to load the pretrain weights.
# for param in model.parameters():
#     param.requires_grad = False
for param in model.parameters():
    param.requires_grad = False

model.avgpool = Identity() # 站位层，使得该层啥事不做
model.classifier = nn.Sequential(nn.Linear(512, 100),
                                 nn.ReLU(),
                                 nn.Linear(100, 10)) # 修改原模型的后几层
model.to(device)


# Load Data
train_dataset = datasets.CIFAR10(
    root="dataset/", 
    train=True, 
    transform=transforms.ToTensor(), 
    download=True)

train_loader = DataLoader(dataset=train_dataset, batch_size=batch_size, shuffle=True)

# Loss and optimizer
criterion = nn.CrossEntropyLoss()
optimizer = optim.Adam(model.parameters(), lr=learning_rate)

# Train Network
for epoch in range(num_epochs):
    losses = []

    for batch_idx, (data, targets) in enumerate(tqdm(train_loader)):
        # Get data to cuda if possible
        data = data.to(device=device)
        targets = targets.to(device=device)

        # forward
        scores = model(data)
        loss = criterion(scores, targets)

        losses.append(loss.item())
        # backward
        optimizer.zero_grad()
        loss.backward()

        # gradient descent or adam step
        optimizer.step()

    print(f"Cost at epoch {epoch} is {sum(losses)/len(losses):.5f}")

# Check accuracy on training & test to see how good our model


def check_accuracy(loader, model):
    if loader.dataset.train:
        print("Checking accuracy on training data")
    else:
        print("Checking accuracy on test data")

    num_correct = 0
    num_samples = 0
    model.eval()

    with torch.no_grad():
        for x, y in loader:
            x = x.to(device=device)
            y = y.to(device=device)

            scores = model(x)
            _, predictions = scores.max(1)
            num_correct += (predictions == y).sum()
            num_samples += predictions.size(0)

        print(
            f"Got {num_correct} / {num_samples} with accuracy {float(num_correct)/float(num_samples)*100:.2f}%"
        )

    model.train()


check_accuracy(train_loader, model)

测试结果

Checking accuracy on training data
Got 29449 / 50000 with accuracy 58.90%

可以看到本次预训练模型的导入，测试结果并不理想。但并不妨碍我们对Pytorch预训练和微调的学习。

参考来源

【1】 https://www.youtube.com/watch?v=qaDe0qQZ5AQ&list=PLhhyoLH6IjfxeoooqP9rhU3HJIAVAJ3Vz&index=8
【2】https://github.com/aladdinpersson/Machine-Learning-Collection/blob/master/ML/Pytorch/Basics/pytorch_pretrain_finetune.py