Reimplementing results for the BIO datasets.

[acharkq]

Thanks for the great work!

I have two questions about re-implementing the results for the BIO datasets since the code is not included in this repo.

• In your paper, does the fine-tuning for BIO use the same hyperparameters as that for CHEM?
• In the original pretrain-gnn code, the code for BIO reports two test results -- test_acc_easy and test_acc_hard . May I know which one do you report in the paper, or do you use the average of both?

Thanks for your time!