分布式图加载log显示的node数随worker数变化而变化是正常的吗？

我在使用分布式图训练时，发现log中显示的图node数随worker数变化而变化，比如100个node的图，1个worker，log显示GraphNode=100，2个worker，log显示GraphNode=50,这样是正常的吗？
看代码实现中graph_builder.cc:147处的逻辑打印出来的node数应该是所有thread之和啊，但是我这边的log看起来貌似只打印了一个worker的图加载数量，这样正常吗？
1个worker

10个worker

Start by inspecting graph_builder.cc around line 147 and compare the GraphNode log output from the one-worker and multi-worker runs shown in the issue. Trace whether the value represents one worker's local graph or an aggregate across workers and threads. Done means the logging behavior and expected node count are established, with any discrepancy clearly identified.

Written by the indexing model from the issue text.