You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
First of all, I would like to thank the authors for open-sourcing the code for the paper. I have some questions regarding the implementation of the coupled graph ODE described in the paper, and I hope the authors can help me.
As I understand it, the paper computes the gradients dX and dE separately using differential equations (4) and (5). However, in the CoupledGraphODE.py file, within the CoupledODEFunc class, it seems that only dX is propagated, while dE is not. Additionally, the implementation uses dX = alpha / 2 * grad_node - input_embedding + grad_node2 - input_embedding, which appears to be somewhat different from equation (5) in the paper.
The text was updated successfully, but these errors were encountered:
First of all, I would like to thank the authors for open-sourcing the code for the paper. I have some questions regarding the implementation of the coupled graph ODE described in the paper, and I hope the authors can help me.
As I understand it, the paper computes the gradients dX and dE separately using differential equations (4) and (5). However, in the CoupledGraphODE.py file, within the CoupledODEFunc class, it seems that only dX is propagated, while dE is not. Additionally, the implementation uses dX = alpha / 2 * grad_node - input_embedding + grad_node2 - input_embedding, which appears to be somewhat different from equation (5) in the paper.
The text was updated successfully, but these errors were encountered: