编写代码，使用pytorch geometric实现GAT处理Cora数据集

在PyTorch Geometric (PyG)库中，Graph Attention Network (GAT)是一种流行的图神经网络模型，用于处理节点分类任务。下面是一个简单的示例，展示了如何使用GAT处理Cora数据集：

# 首先，你需要安装必要的库
import torch
import torch.nn as nn
from torch_geometric.data import DataLoader
from torch_geometric.datasets import Planetoid

# 加载Cora数据集
dataset = Planetoid(root='./data', name='Cora')
train_dataset, test_dataset = dataset[0], dataset[1]

# 数据加载器
train_loader = DataLoader(train_dataset, batch_size=64)
test_loader = DataLoader(test_dataset, batch_size=64)

# 定义GAT模型
class GAT(nn.Module):
    def __init__(self):
        super(GAT, self).__init__()
        self.conv1 = GraphConv(in_channels=dataset.num_features,
                                out_channels=8, heads=8, concat=True)
        self.conv2 = GraphConv(in_channels=8 * 8, out_channels=16, heads=1, concat=False)
        self.fc = nn.Linear(16, dataset.num_classes)

    def forward(self, data):
        x, edge_index = data.x, data.edge_index
        x = F.relu(self.conv1(x, edge_index))
        x = F.dropout(x, p=0.6, training=self.training)
        x = self.conv2(x, edge_index)
        x = F.log_softmax(self.fc(x.mean(dim=1)), dim=1)
        return x

model = GAT()

# 使用交叉熵损失函数和Adam优化器
criterion = nn.CrossEntropyLoss()
optimizer = torch.optim.Adam(model.parameters(), lr=0.005)

# 训练模型
num_epochs = 200
for epoch in range(num_epochs):
    for data in train_loader:
        optimizer.zero_grad()
        out = model(data)
        loss = criterion(out[data.train_mask], data.y[data.train_mask])
        loss.backward()
        optimizer.step()

# 测试模型
model.eval()
with torch.no_grad():
    test_acc = (model(test_dataset)[test_dataset.test_mask] == test_dataset.y[test_dataset.test_mask]).float().mean()
    print(f"Test accuracy: {test_acc.item():.4f}")