Hello,
I am very interested in the GraphBAN project you published and am currently attempting to use GraphBAN for related research. However, I have encountered an issue: When I train the model using the dataset and code provided in the case_study directory, the trained model fails to produce predictions that are identical or similar to your results provided in your paper' s Supplementary Data 1 and Data 2 . Yet, when I use the trained model you provided on Zenodo for direct prediction, the outcomes align with your study.
I would like to ask why this discrepancy might occur. Were the training dataset and hyperparameters you used for model training the same as those provided in the case_study folder? I have reviewed the code I ran and thought about this extensively, but cannot identify the root cause of the problem.
I would truly appreciate any insights you might have. Thank you for your time and response!
Hello,
I am very interested in the GraphBAN project you published and am currently attempting to use GraphBAN for related research. However, I have encountered an issue: When I train the model using the dataset and code provided in the case_study directory, the trained model fails to produce predictions that are identical or similar to your results provided in your paper' s Supplementary Data 1 and Data 2 . Yet, when I use the trained model you provided on Zenodo for direct prediction, the outcomes align with your study.
I would like to ask why this discrepancy might occur. Were the training dataset and hyperparameters you used for model training the same as those provided in the case_study folder? I have reviewed the code I ran and thought about this extensively, but cannot identify the root cause of the problem.
I would truly appreciate any insights you might have. Thank you for your time and response!