Hi! I'm joining in on this late, but I've been int...

# datascience

[mattmoore]

Hi! I'm joining in on this late, but I've been interested in having TensorFlow and something akin to Keras models available in Kotlin as well. I've started a project. Note that I've just very recently begun experimenting. It's very superficial right now, and has a long way to go. However, this concept excites me. I'm working on it in my spare time, but if anyone's interested in working on this with me, I'd be very happy to work with folks on this, and hopefully build something that makes sense. Perhaps if more folks join on the same project this can become a reality? If anyone has a better code base already started, or better ideas, I'm 100% onboard with dumping my very beginning project and starting with something that's further along.

I was also looking into adopting kmath, but have not yet had time to fully vet this out.

My goal is to have this work eventually with both Kotlin JVM as well as Native.

I set a site up for this: https://tensorflow-kotlin.dev

The github repo is here: https://github.com/TensorFlow-Kotlin/tensorflow-kotlin
Thread in Slack Conversation

[Michal Harakal]

Hi @mattmoore. I got really impressed about the project, having nice web page is a good start. I was surprised with storing models in the repo itself, even with using Git LFS. I am not sure about storing big binaries in git. As a solution, I have already implemented gradle downloader “simple one” plugin, in order to download them with gradle tasts. Question is, does it make sense to you to create in MR and contribute a little bit to the project from my side? https://github.com/michalharakal/flagsam/blob/master/buildSrc/src/main/kotlin/de/harakal/flaggie/gradle/DownloadModelTask.kt

[mattmoore]

Hi @Michal Harakal! Yeah right now I have little more than a site and a very simple start at tinkering with the TF API. Long term I won't put models (or other binaries) in the project. I definitely want to have a built-in downloader for models and datasets for both training and running models. I'd be very happy to have others contribute. You're more than welcome to open a pull request to github. I haven't done much with the project recently, but am planning to get back to some of it this weekend. I am working next on a very simple POC to demonstrate TF training in Kotlin, available as an API. Once I have some decent models training through Kotlin, I can start looking at organizing the code a little better. 🙂

[mattmoore]

Also, I created #tensorflow - feel free to join. Hopefully once more of the proof of concept is fleshed out more folks might be interested in joining.

[Michal Harakal]

I’ve just started my DS learning, so I will only contribute to ML part a few moths later, but having some experience with multiplatform and Android I can hopefully the first MR soon …

[mattmoore]

No worries! Having a download util would be very helpful and much appreciated!

[Michal Harakal]

🙂 https://github.com/TensorFlow-Kotlin/tensorflow-kotlin/pull/2

[Michal Harakal]

@mattmoore could we have a short call about your project? I would like to contribute more but I am missing some peaces of informations ...

[Michal Harakal]

@mattmoore I’am still scratching my head around the level of abstraction. I see a huge advantage of dsl and making TF kotlin first. But in the moment you are using jvm library. Do you have some plans to go multiplatform ?

[mattmoore]

@Michal Harakal Hi! Sorry for the late response. Yes, we could do a call.

The goal is to go multiplatform, but there are a few ways to approach it and some complexities with the TF API that I've been thinking through. I'm booked up through Thursday of next week but I could sync up some time after that.

[Michal Harakal]

NP, let me know, you can ping me then with DM here on slack