Gabrijel Boduljak

I think it is a great idea to have support for GNNs in MLX. However, I think that it is better to implement GNN-related stuff in a separate library (e.g....

When it comes to tests, you could test whether neighborhood aggregation is correctly implemented.

> > I think it is a great idea to have support for GNNs in MLX. However, I think that it is better to implement GNN-related stuff in a separate...

It makes sense to close this, given the discussion in https://github.com/ml-explore/mlx/pull/398.

Ideally, we need examples in docs and tests for the binding. I will add some tests.

> The most probably future path, imho, is to expose `scatter_{op}` so I think the work you are doing is not in vain but I am writing so that we...

> As per @angeloskath's suggestion, I will work on exposing `scatter_{op}` instead of generic `scatter`. This is now done. @francescofarina please see https://github.com/francescofarina/mlx/pull/2/.

@awni @angeloskath Could you take a look at this draft implementation?

@angeloskath, do you have any performance concerns regarding pooling implementation using ``as_strided`` ?

@angeloskath I implemented the requested changes. In addition, I separated the generic implementation into ``MaxPooling1d, MaxPooling2d, AvgPooling1d`` and ``AvgPooling2d`` and updated the docs accordingly. However, I am not sure about...