AnacletoLAB / grape

🍇 GRAPE is a Rust/Python Graph Representation Learning library for Predictions and Evaluations

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Question about custom PyG and PyKeen models

Alvaro-Ciudad opened this issue · comments

First of all, congratulations on a great library. I just happen to have some questions surrounding the use of custom models alongside your preprocessing module.
I have a use case that might benefit from the flexibility of implementing custom models in PyG or PyKeen, or using already implemented models for which grape might not have support already. I ve seem that embiggen has some implemented wrappers, and I wonder if you could direct me to an example of how to use those PyG and PyKeen wrappers to implement custom models.
I also think this could be a really cool tutorial to have in your documentation, and I would be ok with contributing to adding support to more PyKeen models if desired :)
Thanks

Oh, ignore the pykeen part, I just found the wrappers for the pykeen library. Still interested in PyG, for the more complex GNNs and cuda support tho :)

Hi @Alvaro-Ciudad! I still have to make that tutorial, I'll work on it later this week. The TLDR of it is to implement the abstract interface for the task that you want to make the model for.

Great, thanks for your quick answer :)