Hello, In section3 COMPUTATIONAL COMPLEXITY ANALYSIS, you claimed the gpu usage of bgrl on Coauthor CS is 2.86 GB, while i follows the discription of read.md. By running python3 train_transductive.py --flagfile=config/coauthor-cs.cfg. The total GPU usage is 5.6 G. what am i missing? Maybe this experiments conducted in anther setting, which is not consistent with here u provided? like a smaller hiddden units? half of them? right
Hello, In section3 COMPUTATIONAL COMPLEXITY ANALYSIS, you claimed the gpu usage of bgrl on Coauthor CS is 2.86 GB, while i follows the discription of read.md. By running python3 train_transductive.py --flagfile=config/coauthor-cs.cfg. The total GPU usage is 5.6 G. what am i missing? Maybe this experiments conducted in anther setting, which is not consistent with here u provided? like a smaller hiddden units? half of them? right