euler有没有支持动态网络embedding的算法？

例如我有一个有向同构图，图本身很大（有上亿条边）但更新的频率比较慢（例如每天增加/删除的边只占所有边数量的不到1%，因此用完整的图重新训练一次开销太大了），请问euler中有没有内置的模型支持每天输入一个增量图，只对增量图中涉及的节点的embedding进行更新的模型呀？

一个问题是每天都会有一些已经之前存在的边被删除，因此不太了解怎么用euler那个json格式表示昨天还存在但今天被删除的边

Read the issue's question about Euler's graph input format and incremental embedding updates; it does not identify a source file, model, or test to start from. Determine whether Euler supports incremental training for changed nodes and how its JSON graph format represents deleted edges; done would mean a documented answer or a scoped implementation, but the issue provides no decision or code pointers.

Written by the indexing model from the issue text.