Adam Gibson

Ah this was the wrong issue. I meant to close the helpers one.

@eix128 are you using the fork? https://github.com/MobiVM/robovm I just want to be clear, since the original robovm was shuttered long ago. @johanvos actually did some ios work a while ago....

We need to be able to reproduce this to help you. This doesn't look related to dl4j from what I can see. These lines: ``` 1 0x00007fb0d637ea78 in abort ()...

We can import fast text from keras though. /cc @maxpumperla

@eix128 it's kind of hard to tell from the web page where to get started for supporting the SDK here: https://www.microej.com/product/sdk/ do you have more info? We already support android...

@jljljl javacpp extracts all binaries on disk to work, but we could look in to making libnd4j run on this pretty easily. It's standalone and just runs math operations. I...

@jljljl @eix128 one of the major issues I see has to do with binary size. We've discovered that the binary size for libnd4j being what it is (~150 MB for...

@jljljl micro sd is how many embedded computers do it. Regarding this, actually no not really. The overhead comes from having a ton of ops built in. A simple thing...

@eix128 we got rid of that long ago. We used to have a pure java backend but found it to be impractical save for very minor use cases.

@braeunlich hmm...this hasn't happened specifically on my machine and wasn't an issue in our internal testing. We currently build using this GH actions setup: https://github.com/eclipse/deeplearning4j/blob/master/.github/workflows/build-deploy-windows.yml - it's hard for me...