More languages to support

[alexander-nsk]

Hello :)
It is possible to use MLEM models not only with Python but also with other languages?
Thank you.

[aguschin]

Hi @alexander-nsk! What language do you mean exactly? Could you please describe your use case in more details?

[denis-akvelon]

Hi, @alexander-nsk!
We are working on it, look at https://github.com/akvelon/MLEM-SDK-for-Java and https://github.com/akvelon/MLEM-SDK-for-C-Sharp

[aguschin]

cc @mike0sv

[alexander-nsk]

Hi @alexander-nsk! What language do you mean exactly? Could you please describe your use case in more details?

For example, in case I want to provide MLEM functionality to my Android application based on Java/Kotlin.
Thank you.

[alexander-nsk]

Hi, @alexander-nsk! We are working on it, look at https://github.com/akvelon/MLEM-SDK-for-Java and https://github.com/akvelon/MLEM-SDK-for-C-Sharp

I'll check it.
Thank you.

[aguschin]

For example, in case I want to provide MLEM functionality to my Android application based on Java/Kotlin.

@alexander-nsk, could you please clarify what MLEM functionality do you mean exactly? From what you say I can guess you want to

1. Train model in Python, save it with MLEM
2. Load MLEM model in Java/Kotlin, pass data to it and generate predictions

Am I right, or you have something else in mind?

[alexander-nsk]

For example, in case I want to provide MLEM functionality to my Android application based on Java/Kotlin.

@alexander-nsk, could you please clarify what MLEM functionality do you mean exactly? From what you say I can guess you want to

1. Train model in Python, save it with MLEM
2. Load MLEM model in Java/Kotlin, pass data to it and generate predictions

Am I right, or you have something else in mind?

Yes, You are right 👍
Thank you.

[aguschin]

Thanks @alexander-nsk! We're thinking about this use case. Specifically, we think about implementing export to onnx format, since you can use the model in onnx format in most languages. Will it work for you?

[aguschin]

Hi @alexander-nsk! A friendly ping :) WDYT about my last question?

[alexander-nsk]

Hi @alexander-nsk! A friendly ping :) WDYT about my last question?
Hello, @aguschin .

I found some information about onnx format with Java. It seems like a very low-level approach, not exactly what I would expect from a framework designed to ease cross-platform model sharing.
Thank you.

[aguschin]

Ok, got it. Would be happy to learn if there are more friendly/helpful formats to export models into. Since MLEM saves models in Python, I guess exporting models to cross-platform formats is the easiest way to use them in another languages.