search

google-deepmind / graph_nets

Build Graph Nets in Tensorflow
https://arxiv.org/abs/1806.01261
Apache License 2.0
5.34k stars 783 forks source link

Is this project still live? #148

Open caiodadauto opened 2 years ago

caiodadauto commented 2 years ago

Hey, in the last 2 years I've been working with graph_nets in my work (specially in my PhD). I even implement new blocks to work with recurrent layers and other minor features (I didn't send a pull request yet, because I cannot find time to organize the code and implement the tests).

Anyway, there are not any commit in the last 2 years, is there any plan to continue this project?

If graph_nets will not be continue, do you have any advice regarding other frameworks? I am not a big fan of Pytorch, I really like to use Tensorflow and Sonnet, then I am thinking to try DGL. However, I am not sure if I have the same degree of freedom that graph_nets gives to me.

8bitmp3 commented 2 years ago

Can't speak for the team but have you seen https://github.com/deepmind/jraph? (copying @jg8610).

alvarosg commented 2 years ago

Thanks for your message @caiodadauto, the library itself is not deprecated but our team shifted to JAX, and as @8bitmp3 mentioned we have also developed a JAX version of the library. That said if you have any contributions for the graph_nets library that are general purpose and you would like to provide, and are happy to clean up the code and add tests, we would be happy to review it and merge it (although before doing the work, perhaps create an issue/feature request for each piece of functionality so we can discuss it!).

We have also advised on the design of TF-GNN which is a new tensorflow library for GNNs and provides similar functionality to the graph_nets library, but also offers native support for graphs with explicitly typed nodes and edges, and is designed around Keras, which is one of the most popular model frameworks for TensorFlow. If you are interested on exploring this, a good starting point would be the shortest path demo which we migrated to TF-GNN.