rand_indx = torch.tensor(np.random.choice(np.arange(adj_label.shape[0]), batch_size)).type(torch.long) ？ what is adj_label？

yanghu819 / Graph-MLP

73 stars 21 forks source link

Closed ghost closed 3 years ago

ghost commented 3 years ago

rand_indx = torch.tensor(np.random.choice(np.arange(adj_label.shape[0]), batch_size)).type(torch.long) 训练数据包括验集数据和测试集数据？

yanghu819 commented 3 years ago

adj_label是adj的邻接信息不是ground truth，没有用到val or test的node class label。

yanghu819 commented 3 years ago

麻烦你看明白transductive node classification的设定，用全部的Adj和train 的node feature作为train，建议参考tkipf的代码搞明白基础