tencentyouturesearch / hig-graphclassification Goto Github PK

Hi authors,
Thank you for your great work! I noticed that the result of 'the mean of the 4 cards ap' is different from the result of 'gather all pred and labels of different cards and do evaluation once'. And the latter method's result tends to be lower than the former one. It seems that when doing evaluation, Graphormer is using the former method. May I know that if you have the valid and test result of Graphormer model evulating the whole dataset once? Thank you!

Hi, thanks for your contribution. I have a question in the implementation process. For molcpba, firstly we use graphormer to pretrain on CQM4Mv2 , which in num_class is 1. Thus, the size is torch.Size([1, 1024]). However, when we finetune on the molpcba, the num_class is 128，size is torch.Size([128, 1024]). Thus their size is mismatch. How do we address it? Thanks.

Hi,
Where is the pretrained model for ogbg-molpcba?