Graph Classification Accuracy Calculation?

[gdreiman-insitro]

Hi thanks for the very interesting work.

I am wondering about how the metrics for the graph classification were calculated. In the paper, Table 2 is labeled at "Accuracy (%)". In both the DGL and pyg versions of this repo, it seems like F1 is calculated but reported as accuracy?

DGL

evaluate_graph_embeddings_using_svm calculates F1 here graph_classification_evaluation returns the return of evaluate_graph_embeddings_using_svm (test_f1) here test_f1 is appended to acc_list here final_acc is the mean of acc_list here

pyg

evaluate_graph_embeddings_using_svm calculates F1 here graph_classification_evaluation returns the return of evaluate_graph_embeddings_using_svm (test_f1) here test_f1 is appended to acc_list here final_acc is the mean of acc_list here

Is the code in the repo what was used to generate the results in the paper? I am having trouble replicating some results and wondering if the results in the paper are accuracy or F1.

Thank you