# SimpleChains.jl extension to Graph Neural Networks
**URL:** https://discourse.julialang.org/t/simplechains-jl-extension-to-graph-neural-networks/80422
**Category:** Machine Learning
**Tags:** flux, graphneuralnetworks
**Created:** [May 3, 2022, 12:29pm UTC](https://discourse.julialang.org/t/simplechains-jl-extension-to-graph-neural-networks/80422 "2022-05-03T12:29:49Z")
**Posts on this page:** 3
**Page:** 1
### Author:  [@DoktorMike](https://discourse.julialang.org/u/DoktorMike)
#### Post date: [May 3, 2022, 12:29pm UTC](https://discourse.julialang.org/t/simplechains-jl-extension-to-graph-neural-networks/80422/1 "2022-05-03T12:29:49Z")
So yesterday I attended the Copenhagen julia meetup and saw a new Neural network toolbox called `SimpleChains.jl` which aims to be fast for small neural networks basically. Often Graph Neural Networks are also quite small in number of parameters and I was wondering if there are lessons from SimpleChains we can use to make our Graph Neural Network implementations faster? I’m thinking primarily of `GeometricFlux.jl` and `GraphNeuralNetworks.jl`. To me it seems as though they are based directly on Flux and as such will suffer from the same “small network” performance issues as `SimpleChains.jl` aims to solve. Not sure if I’m right about this and happy to be corrected.
Thoughts?
---
### Author:  [@ChrisRackauckas](https://discourse.julialang.org/u/ChrisRackauckas)
#### Post date: [May 3, 2022, 2:41pm UTC](https://discourse.julialang.org/t/simplechains-jl-extension-to-graph-neural-networks/80422/2 "2022-05-03T14:41:39Z")
Open an issue. I think small graphs are a great use case for this, since indeed exploiting their sparsity makes them memory bound and good on CPUs. Let’s discuss with @Elrod
---
### Author:  [@Elrod](https://discourse.julialang.org/u/Elrod)
#### Post date: [May 4, 2022, 1:25am UTC](https://discourse.julialang.org/t/simplechains-jl-extension-to-graph-neural-networks/80422/3 "2022-05-04T01:25:47Z")
Sure, PRs are welcome.
I can also add more documentation describing an interface, but it may be helpful to flesh it out, especially with guidance from folks who have more experience doing ML work than I do.