pysyft/client.py

import syft as sy
import torch
import torch.nn as nn
import torch.optim as optim

class SimpleCNN(nn.Module):
    def __init__(self):
        super(SimpleCNN, self).__init__()
        self.conv1 = nn.Conv2d(3, 6, 5)
        self.pool = nn.MaxPool2d(2, 2)
        self.conv2 = nn.Conv2d(6, 16, 5)
        self.fc1 = nn.Linear(16 * 5 * 5, 120)
        self.fc2 = nn.Linear(120, 84)
        self.fc3 = nn.Linear(84, 10)

    def forward(self, x):
        x = self.pool(torch.relu(self.conv1(x)))
        x = self.pool(torch.relu(self.conv2(x)))
        x = x.view(-1, 16 * 5 * 5)
        x = torch.relu(self.fc1(x))
        x = torch.relu(self.fc2(x))
        x = self.fc3(x)
        return x

def train_cnn_model(images, labels):
    # 创建模型
    model = SimpleCNN()
    criterion = nn.CrossEntropyLoss()
    optimizer = optim.SGD(model.parameters(), lr=0.001, momentum=0.9)

    # 训练模型
    print("Starting training...")
    for epoch in range(2):  # 只训练2个epoch作为示例
        running_loss = 0.0
        for i in range(len(images)):
            # 获取数据
            inputs = images[i].unsqueeze(0)
            target = labels[i].unsqueeze(0)

            # 前向传播
            optimizer.zero_grad()
            outputs = model(inputs)
            loss = criterion(outputs, target)

            # 反向传播
            loss.backward()
            optimizer.step()

            running_loss += loss.item()
            if i % 10 == 9:  # 每10个batch打印一次
                print(f'[{epoch + 1}, {i + 1:5d}] loss: {running_loss / 10:.3f}')
                running_loss = 0.0

    return model

def main():
    # 直接连接到已运行的服务器
    client = sy.login(
        url="localhost:8093",
        email="researcher@cifar10.research",
        password="syftrocks"
    )

    try:
        # 获取数据集
        dataset = client.datasets["CIFAR10 Training Dataset"]
        print(f"Retrieved dataset: {dataset.name}")

        # 获取资产
        images, labels = dataset.assets
        print(f"Retrieved assets: {images.name}, {labels.name}")

        # 使用模拟数据
        train_images = images.mock
        train_labels = labels.mock

        # 创建项目
        project_description = """
        The purpose of this study is to train a CNN model on CIFAR10 data.
        The model architecture includes two convolutional layers and three fully connected layers.
        We will evaluate the model's performance on the training data.
        """
        
        research_project = client.create_project(
            name="CIFAR10 CNN Project",
            description=project_description,
            user_email_address="researcher@cifar10.research"
        )

        # 创建远程函数
        remote_train_function = sy.syft_function_single_use(
            images=train_images,
            labels=train_labels
        )(train_cnn_model)

        # 创建代码请求
        code_request = research_project.create_code_request(
            remote_train_function,
            client
        )

        # 等待数据所有者批准请求
        print("Waiting for data owner to approve the request...")
        input("Press Enter after the data owner has approved the request...")

        # 执行训练
        model = client.code.train_cnn_model(
            images=train_images,
            labels=train_labels
        )

        print('Training completed successfully!')

    except Exception as e:
        print(f"Error occurred: {str(e)}")

if __name__ == '__main__':
    main()