Thank you for providing the Pronet code. I have reproduced Pronet experiments multiple times under identical conditions, including using the same graph cards. However, there is a noticeable performance gap (approximately 1.5-5%) on both ProFunct and HomologyTAPE datasets between my reproduced results and those reported. Could you kindly share the optimal hyperparameter settings, rather than the default ones?
Thank you for providing the Pronet code. I have reproduced Pronet experiments multiple times under identical conditions, including using the same graph cards. However, there is a noticeable performance gap (approximately 1.5-5%) on both ProFunct and HomologyTAPE datasets between my reproduced results and those reported. Could you kindly share the optimal hyperparameter settings, rather than the default ones?