allenhaozhu

Sorry to confuse you, you could check issues. I give different ways for this. Tips: SGC and SSGC are very sensitive to weights decays.......

In issue#13, I give a initialization for 83.5% in cora. Like what I explain, after fixed the bug, the performance slightly drop but the bug only on the cora,pubmed and...

Thank you very much for your help! You have no idea what's a big favour for me. Let me answer some questions. Yes, this code has bugs about the inplace...

> Hi, allenhaozhu. > > I have read your paper and the code, but I found that the SSGC computation is not exactly right, where `emb += features` is an...

> add .item() on validation acc and loss > > solve the Runtime error below: > `RuntimeError: Expected all tensors to be on the same device, but found at least...

In our paper, to demonstrate an APPNP similar form, we add X for every propagation. However, it is easy to move X outside the sum function and then add X...

> In your paper,  Eq. 13 indicates: embedds =  a_X + (1-a)(TX+T^{2}X+....); i.e., you add X only once? But in the code function "sgc_precompute", embedds = a_X +(1-a)(X+TX+T^{2}X+....);  i.e., you...

> Hi, thanks for the code! > > Do I misunderstand the code? Have you raised the power to the `(1 - alpha)`? in which case, embeds = ax +...

> Maybe I miss some information, or I made some mistakes here. > > Are we supposed to do `ax + (1-a)* 1/k *(T x + T^2 Tx ...T^k x)`...

I guess the difference is not a issue for the final result. Because it is more like a global scale (very close to 1) to the final features. My fault.