lucfra / LDS-GNN

Learning Discrete Structures for Graph Neural Networks (TensorFlow implementation)
Other
192 stars 40 forks source link

GPU implementation #10

Closed adhikarirsr closed 2 years ago

adhikarirsr commented 3 years ago

This is such an interesting idea and nicely done work. Thanks a lot!

Is there a GPU implementation of your algorithm? I was testing digits example and it takes a lot of time.

lucfra commented 2 years ago

HI @adhikarirsr ,

Sorry for the late reply and thanks for your interest in our work!

If I remember well, the code should run on the GPU if it is available. Probably it's slow because the runtime scales quadratically in the number of training points. A way to go around this could be to implement a stochastic version of the algorithm, but this is not in our plans at the moment. See also this https://github.com/lucfra/LDS-GNN/issues/6

Hope this helps, and let me know if there are any other issues.

Cheers Luca

adhikarirsr commented 2 years ago

Hi @lucfra ,

I figured out that it is GPU compatible. That's why closed the issue. I had to install tf 1.15 but while doing so I installed CPU version of the tf. That's why I was having the issue. I was just being a fool. I have a few more questions that are not issues. Is it fine if I email you @ igor_mio@hotmail.it ?

Thanks for the reply.

lucfra commented 2 years ago

Sure, please go ahead (I'm a bit busy these days but will try to reply soon!)

adhikarirsr commented 2 years ago

Did you get my email?