Question about the output of UGraphEmb for PTC dataset

[annitrolla]

I run the UGraphEmb without any modifications for PTC dataset and successfully obtained the results. However, I have a question about the output of your model. Namely, in the logs folder test_info.klepto contains K_graph_embs_mat, which I understand to be graph embeddings of the test, with its shape being (344, 256) - 344 graphs with 256 embedding space. The problem is that the loaded test dataset contains only 69 graphs. Therefore, it seems that it contains all of the graphs. So my question is whether the resulting embeddings are sorted by gids or how to map them to their corresponding graph indices?

Thanks for your help and an interesting work!