Hello, your work is interesting and inspires me a lot!
But recently when running your code in this project, I am a bit confused.
When I directly execute commands you provided in demo.txt, there is a decrease in performance compared to the results in your paper, especially in the PubMed, CoraFull, and Computer datasets.
May I ask if there are some other training tricks, or can you provide commands with better hyperparameters in demo.txt for achieving performance in your paper? Thanks.