搭建一个基于pytorch和DGL的gcn模型，并给出完整代码

在 PyTorch 中搭建一个基于 DGL 的 GCN 模型，需要以下步骤:

1. 安装 DGL 库: `pip install dgl`
2. 导入 DGL 中 GCN 模型所需的库:

import dgl
import dgl.function as fn
import torch
import torch.nn as nn

3. 定义 GCNLayer 类继承自 nn.Module，并实现 forward() 函数:

class GCNLayer(nn.Module):
    def __init__(self, in_feats, out_feats):
        super(GCNLayer, self).__init__()
        self.linear = nn.Linear(in_feats, out_feats)

    def forward(self, g, inputs):
        # g is the graph and the inputs is the input node features
        # first set the node features
        g.ndata['h'] = inputs
        # then send the node features to its neighbors
        g.update_all(fn.copy_src('h', 'm'), fn.sum('m', 'h'))
        # apply linear transformation
        h = self.linear(g.ndata['h'])
        return h

4. 构建 GCN 模型，并定义损失函数和优化器:

class GCN(nn.Module):
    def __init__(self, in_feats, hidden_size, num_classes):
        super(GCN, self).__init__()
        self.layer1 = GCNLayer(in_feats, hidden_size)
        self.layer2 = GCNLayer(hidden_size, num_classes)

    def forward(self, g, inputs):
        h = self.layer1(g, inputs)
        h = torch.relu(h)
        h = self.layer2(g, h)
        return h

# define loss function and optimizer
criterion = nn.CrossEntropyLoss()
optimizer = torch.optim.Adam(model.parameters(), lr=0.01)

5. 加载数据并训练模型:

# load your data
g, inputs, labels = ...

# training loop
for epoch in range(100):
    logits = model(g, inputs)
    loss = criterion(logits, labels)

    optimizer.zero_grad()
    loss.backward()
    optimizer.step()

以上是一个简单的 GCN 模型的实现，你可以根据你的数据和需求来修改