Alexander comments

Results 7 comments of


                                            Alexander

Node level learning?

Hi @vcanogil, yes you can :) Just exclude readout and add labels (y) corresponding to each node (y.shape[0] = x.node_feature.shape[0])

Cannot save model as specified in the documentation

Thanks for the observation @RaphaelRobidas, I will check it out.

Cannot save model as specified in the documentation

It seems to be an issue related to the ExtensionType API of TF. The 'auxiliary' field of the GraphTensor is a `Mapping[str, tf.Tensor]` type which internally creates a `ImmutableMapping`, which...

Cannot save model as specified in the documentation

@RaphaelRobidas Okay interesting, good to know. And good to know you could save your models eventually, although not in .keras format.

Cannot save model as specified in the documentation

Btw, I would like to migrate to Keras 3 (and TF>=2.16); however, Keras 3 does not yet support [extension types](https://www.tensorflow.org/guide/extension_type).

Cannot save model as specified in the documentation

> It seems to be an issue related to the ExtensionType API of TF. The 'auxiliary' field of the GraphTensor is a `Mapping[str, tf.Tensor]` type which internally creates a `ImmutableMapping`,...

Parallelize training in multiple GPUs.

Thanks for the feedback @soulios, I'll look into it :) Is it only DMPNN that is significantly slower? What about e.g., GINConv, GATv2Conv, or perhaps MPNN? And is it specifically...