train_node_type 默认是0 表示只训练第0类顶点的embedding吗

tf.flags.DEFINE_integer('train_node_type', 0, 'Node type of training set.')
train_node_type 默认是0 表示只训练第0类顶点的embedding，
但是采样是在全部类型的节点上游走，采样，这样理解对吗？

Start with the tf.flags.DEFINE_integer declaration for train_node_type and trace where that flag is consumed in the embedding-training and node-sampling paths. Confirm whether the training set is restricted to node type 0 while walks sample all node types; done means the behavior is clearly explained or the discrepancy is documented for follow-up.

Written by the indexing model from the issue text.