How to handle unseen nodes in link prediction task?

[NiDHanWang]

I've built a link predication model with torch_geometric.nn.SAGEConv:

class SAGE(Module):
    def __init__(self, in_channels: int, hidden_channels: int, out_channels: int, dropout: float = 0.2) -> None:
        super().__init__()
        self.conv1 = SAGEConv(in_channels, hidden_channels)
        self.conv2 = SAGEConv(hidden_channels, out_channels)
        self.dropout = dropout
        self.lin  = Linear(out_channels * 2, 1)

    def forward(self, x: Tensor, edge_index: Tensor) -> Tensor:
        x = F.relu(self.conv1(x, edge_index))
        x = F.dropout(x, self.dropout, training=self.training)
        x = self.conv2(x, edge_index)
        
        return x

During inference, there're could be unseen nodes coming in, how to handle this?